Intense investigation of epithelial cell proliferation in the mammalian intestinal mucosa has resulted in a significant progress in understanding regulatory mechanisms governing cell dynamics in this rapidly self-renewing cell population. This understanding greatly assisted in developing theoretical models of human colorectal carcinogenesis addressing both its molecular mechanisms1, 2, 3, 4 and sequences of events in tumour morphogenesis.5, 6, 7, 8, 9, 10 As numerous studies in this area were devoted to the investigation of colonocyte proliferation and differentiation, relatively little remained known about the final natural destiny of these cells. Obligatory exfoliation of terminally differentiated colonocytes was traditionally believed to be the predominant way of cell loss in colonic epithelium,11 but this popular notion remained essentially hypothetical in the absence of convincing firm evidence. Given that it has become generally admitted that exfoliated colonocyte analysis may open new approaches to colorectal cancer screening and early diagnosis, detailed knowledge and better understanding of the exfoliation of colonic epithelium and its changes in neoplasia is of high practical relevance. The present brief review addresses this surprisingly obscure area with the purpose of clarifying a few points still provoking controversy and confusion. Perspectives of using the phenomenon of colonic cell exfoliation for screening and diagnosis of colorectal cancer and related diseases are then re-evaluated in view of recent advances in the field.
Colonocyte exfoliation in the human colon constitutes a unique mechanism of cell population control that can undergo significant changes under different physiological and pathological conditions. Being closely related to the apoptosis and anoikis, cell exfoliation from colonic epithelium appears to be a relatively rare event in normal conditions, but its rate dramatically increases in neoplasia, when cell removal by apoptosis in situ does not function properly. Several studies show that significant numbers of exfoliated colonocytes are not lost in the faecal contents of the gut, but retained in the mucocellular layer overlying colonic mucosa. Recent observations allow hypothesizing that the mucocellular layer containing exfoliated colonocytes may gradually migrate distally, eventually leading to the accumulation of the cells exfoliated from malignant colorectal tumours on the surface of the rectal mucosa. Implications of exfoliated colonocyte analysis to colorectal cancer screening and early diagnosis are discussed. © 2007 Wiley-Liss, Inc.
Cell exfoliation in the colonic epithelium: Its role in normal physiological conditions and neoplasia
Colonic epithelium is known to be one of the most dynamic cell populations of the human organism.4, 8, 11, 12 Meticulous investigation of the well-structured organisation of the colonic mucosa allowed constructing a detailed and credible model of colonocyte proliferation, differentiation and migration especially important in relation to colorectal carcinogenesis. It is generally accepted that in normal physiological conditions stem cells of colonic epithelium are located at the base of the crypt.4, 7, 8, 12 Their progeny migrates upward, gradually undergoing differentiation and loosing proliferative capacity until terminally differentiated nondividing colonocytes reach surface (luminal) epithelium.4, 11, 12 There these cells end their short lives, being replaced by next generations of differentiating counterparts. Indeed, extremely high colonocyte proliferation rate and constant flow of new cells migrating from crypts toward the lumen demand an adequate efficiency of cell elimination. For decades it was presumed that all highly differentiated colonocytes of the luminal compartment of colonic epithelium are eventually exfoliated into the lumen.7, 11 This theory may be generally correct for the colonic epithelium of rodents,13 but there are good reasons for doubts about its applicability to humans. Although some authors still believe that massive exfoliation is constantly going on with millions of colonocytes shed into the faecal stream daily,14, 15 there is accumulating evidence that the rate of cell exfoliation from normal human colonic mucosa is much lower than it was previously believed.16, 17, 18, 19 Apoptosis in situ followed by the engulfment of apoptotic cells by adjacent colonocytes or subepithelial macrophages appears to be the main pathway of cell death in healthy colonic epithelium,16, 17, 20, 21 colonocyte exfoliation being an important, but auxiliary mechanism of cell loss. This situation can, however, be completely reversed in neoplastic growth. Despite the absence of a complete consensus on the precise order of the emergence of initial neoplastic lesions with both “bottom-up”6, 7, 8, 9, 10 and “top-down”5, 7 theories in use, it is apparent that intercryptal (luminal) epithelium becomes involved in the process very early. Although even small neoplastic loci can bear severe alterations in the homeostatic control of cell reproduction and loss, little is known about colonocyte exfoliation in early tumours. Later exfoliation changes are much better documented. It is well established that profound deregulation of apoptosis is a characteristic feature of cancer.2, 22, 23 Moreover, the loss of cell adhesion associated with malignant progression and normally regarded in relation to the metastatic process24, 25, 26 can further change the pattern of cell loss in a growing tumour in favour of exfoliation. Indeed, a dramatic increase of colonocyte exfoliation in colorectal cancer patients has been reported by several groups.18, 27, 28, 29, 30, 31 Ahlquist et al. observed accumulation of exfoliated malignant cells in the mucocellular layer overlaying tumour surface, but not over unchanged colonic mucosa.18 In a recent study32 we collected exfoliated cells from mucosal surface of resected colon specimens immediately after operations and were able to show that thousands of cells are exfoliated from the surface of malignant colorectal tumours. These cells as well as multicellular fragments of well-preserved neoplastic tissue (Fig. 1) were found within the mucocellular layer not only over tumours, but even at significant distances distally from tumours indicating that exfoliated material tends to migrate alongside the faecal flow, but mostly without being incorporated into the faeces. Even mitotic cells (Fig. 1) could be occasionally detected in the exfoliated clusters of malignant cells, clearly indicating that these cells preserved viability and in some cases proliferative potential. The findings were perfectly in line with clinical reports describing secondary distal, especially anal, tumours in colorectal cancer patients.33, 34, 35 Such tumours, albeit rare, appear to result from an unusual metastatic process involving the reimplantation of viable exfoliated malignant cells migrating within the mucocellular layer. Likewise, occasional intraperitoneal seeding of exfoliated cancer cells during surgery for colorectal cancer can lead to peritoneal carcinomatosis.36
Analysis of the available literature combined with recent observations of our group allows proposing a partially hypothetical scheme of changes in colonocyte exfoliation occurring alongside the growth of malignant tumours (Fig. 2). The scheme highlights the following 3 hallmarks characterizing cell exfoliation in colorectal cancer:
- aextremely high level of cell exfoliation from the surface of malignant tumours;
- bpreferential location of exfoliated cells and their remnants in the mucocellular layer between colonic mucosa and faecal contents of the gut; and
- cgradual distal movement of the cell-containing mucocellular layer.
The first two of these points have already been discussed in a recent review.21 Observations of our group supporting the last point are still awaiting peer-reviewed publications; thus, the author has to consider it as a new hypothesis.
It should be admitted that little is known about cell exfoliation at early stages of carcinogenesis in the human colon. Further studies are needed to reveal molecular mechanisms triggering the dramatic increase of colonocyte exfoliation as well as the precise moment when this “switch” occurs. It seems to be logical that the exfoliation increase coincides with apoptosis deregulation and/or cell adhesion loss, but firm evidence is still absent. It also remains unclear whether cell exfoliation from the surface of colorectal carcinomas should be regarded as a protective mechanism or a manifestation of increasing tumour invasiveness. One can suggest that the enhancement of exfoliation is initially a protective response directed to activating anoikis37, 38 by cell detachment from the basal membrane when apoptosis in situ fails to occur. Alternatively, it may be just another manifestation of neoplastic progression when tumour starts producing autonomous potentially metastatic cells ready for secondary implantation. The latter assumption is corroborated by the cited reports of secondary anal metastases after removal of primary colorectal cancers33, 34, 35 or peritoneal cancer dissemination during operations.36 However, the 2 mechanisms are likely to be inseparably entangled since well-preserved colonocytes, apoptotic cells and cell debris are all abundant in the mucocellular layer of cancer patients.
It should be accepted that even phenomenology of cell exfoliation in colorectal cancer is still far from being completely clear. The scheme in the Figure 2 may be correct till some point in the progression of cancer (e.g., Fig. 2d), but it is impossible to exclude a secondary decrease in cell exfoliation from the primary tumour associated with the development and growth of its local and remote metastases. This is another obscure and theoretically challenging area remaining to be studied.
Methodological approaches to the isolation and analysis of human exfoliated colonocytes
The possibility of obtaining well-preserved epithelial cells from colorectal washings and their use for cytological diagnosis of neoplastic conditions had been described for the first time over 50 years ago.27 However, the proposed procedures of colorectal irrigation and material collection27, 28 looked complicated and were difficult to standardize. This, together with the exclusively cytological approach to colonocyte analysis, resulted in a slow progress in the development of exfoliated cell collection techniques.
The interest to cell exfoliation in the human colon had re-emerged with the introduction of the Vogelstein's genetic model for colorectal carcinogenesis.1 As analytical methods for the detection of cancer-associated molecular markers became available, it was logical to suggest that cells exfoliated from colorectal tumours could provide sufficient amounts of DNA for such analyses. The idea of obligatory colonocyte shedding into the faecal stream discussed in the first section of this review was generally accepted at the time, so it looked transparent that human stool samples could provide exfoliated colonocytes for either cytological characterization or nucleic acid and protein isolation for molecular biomarker analysis. This noninvasive approach to cellular material collection promised to open a convenient way to colorectal cancer early diagnosis and screening.
In the beginning of the nineties, Nair et al. published 2 papers describing exfoliated colonocyte isolation from dispersed samples of human stool by Percoll density gradient centrifugation. The group claimed that millions of viable exfoliated colonocytes could be obtained from a few grams of stool.39, 40 Although those publications certainly had a positive effect attracting attention to an important physiological phenomenon, the main claim of the authors on the isolation of millions of viable colonocytes from small amounts of faecal material was not convincingly substantiated morphologically and remained debatable.
Several groups of scientists attempted exfoliated colonocyte collection from either stool samples or colorectal mucosa (see Table I). It is apparent that attempts of using human faeces as the source of exfoliated colonocytes were only partially successful. Indeed, it is difficult to expect finding viable colonocytes in stool homogenates even if they were entering the faecal milieu in millions. The likelihood of the presence of well-preserved colonocytes in the anaerobic faecal environment rich in bile acids and other cytolytic agents appears to be very low. One possible minor source of cells inside faecal samples can be provided by strands of colonocyte-impregnated mucus from the mucocellular layer that are folded into the stool.21 The latter assumption remains to be proven, but fragments of the mucocellular layer are certainly excreted on the stool surface. In our earlier studies we explored this phenomenon initially in a rat model of colon carcinogenesis,48 and then developed a method for human exfoliated colonocyte isolation based on the use of surface washes from whole stool samples in combination with immunomagnetic bead-based cell separation.19, 30 The method allowed isolation of well-preserved colonocytes and was later employed by other groups for the identification of cancer-associated cellular markers49, 50; however, few colonocytes were usually obtained, whereas squamous cells of the anal canal epithelium often dominated the pool of isolated cells. In addition, the standardization of the technique was extremely difficult because of the necessity to use intrinsically variable whole stool samples. Two other groups also tried immunomagnetic cell capture, using stool homogenates.42, 43 Although application of diagnostic cytology was attempted by one of these groups upon collection of material from cancer patients, no information on cell yield was provided, and cytology results were inconclusive in most cases.43 It is apparent that obtaining material for DNA or RNA examination was the main goal of exfoliated cell isolation in most studies of this type.41, 42, 43
|Cell collection sources||Method of cell isolation||Exfoliated cell yield||Exfoliated cell morphology and squamous (anal) epithelium presence||Faecal contamination of collected material||Technical (reproducibility, standardisation) problems||Applicability of analytical approaches (quantitative/qualitative)||References|
|Small amounts of faecal material (1 g).||Stool dispersion followed by density gradient centrifugation.||Claimed to be millions of viable cells per gram of stool; however, this claim seems to be doubtful.||No convincing evidence of well-preserved cells. The presence of cell components including those derived from squamous epithelium was likely.||Always strong.||The technique appeared to be standardized, but claimed results look exaggerated and not substantiated by cytological analysis, making reproducibility highly questionable.||Isolated material may contain some cells and cell components; thus, biochemical and molecular methods can be applied. Quantitative approaches questionable.||39,40|
|Small amounts of faecal material (5 g).||Stool dispersion followed by a Percoll centrifugation-based technique (a modification of the technique of Nair et al.39, 40).||Appears to be low, however, no estimate given.||Cell debris rather than cells mostly obtained. The presence of squamous cell components was likely.||Always strong.||The technique was found to be difficult to standardize and reproduce. Later abandoned.||RNA isolation and gene expression analysis shown to be possible.||41|
|Small amounts of faecal material (1 g).||Stool dispersion followed by either a Percoll centrifugation-based technique (a modification of the technique of Nair et al.39, 40) or immunomagnetic bead-based separation.||Not assessed.||Isolated cells were not cytologically characterised. The presence of squamous cell components was likely.||Always strong.||Immunomagnetic separation was found to be superior in terms of reproducibility compared to Percoll-based isolation.||Cell isolation did not improve qualities of isolated DNA and RNA compared to direct isolation from faeces.||42|
|Small amounts of faecal material (5–10 g).||Stool homogenisation followed by immunomagnetic bead-based separation.||Not assessed.||Isolation of colonocytes reported to be cytologically confirmed. The presence of squamous cells not evaluated.||Always strong.||Immunomagnetic separation was found to be far superior in terms of reproducibility compared with Percoll-based isolation.||DNA extracted from isolated cells was suitable for amplification and molecular analysis.||43|
|Stool surface (whole stool samples).||Stool surface (part of the mucocellular layer excreted with faeces) was washed with a cell-dispersing solution. Cells isolated using immunomagnetic bead-based separation.||Very low in normal conditions, higher (hundreds) in samples from colorectal cancer patients.||Few well-preserved colonocytes could be isolated. Squamous epithelial cells were commonly present.||Usually strong.||The technique was reproducible; however, its standardization was hardly possible.||Isolated cells could be used for cytological and immunohisto-chemical assessment. DNA isolated from cells could be quantified and successfully amplified.||19,30|
|Techniques based upon cell collection from colorectal mucosa|
|Surface of the colorectal mucosa (distal segment).||Colorectal lavage followed by simple centrifugation (or gradient centrifugation44,45).||High (hundreds to thousands easily identifiable cells).||Well-preserved colonocytes can be isolated. Squamous cells always present.||Variable.||Standardization difficult (variability of the recovered lavage liquid volume).||Standardization problems make quantitation difficult. Cell-based and molecular qualitative analytical methods applicable.||27,28,44,45|
|Surface of the rectal mucosa||Direct collection of exfoliated cells from the mucocellular layer overlaying the surface of rectal mucosa using a thin inflatable membrane.||High (hundreds in normal conditions; thousands in cancer cases or severe inflammatory conditions).||Very well-preserved colonocytes. Occasional squamous epithelium presence.||Mostly low (strong contamination in less than 10% cases).||Well standardized and reproducible.||Collected cells can be used for a wide range of quantitative and qualitative analytical techniques.||32,46,47, A. Loktionov et al., unpublished observations|
The analysis of the reviewed reports on colonocyte isolation from human stool samples shows that none of the employed techniques could provide significant amounts of exfoliated cells suitable for reliable cytological assessment or intracellular biomarker identification. Given that recent efforts focused on the direct extraction of human DNA and RNA from stool samples have produced encouraging results (see later), the idea of preliminary isolation of exfoliated colonocytes from stool with the same only purpose of extracting nucleic acids now starts to look less attractive.
It was known for decades that exfoliated colonocytes could be obtained by colorectal lavage (see Table I); however, this procedure, being inconvenient and difficult to standardize, is now rarely employed. Nevertheless, it appeared that direct collection of exfoliated colonocytes from the surface of human rectal mucosa could provide substantial amounts of well-preserved cells. Following the hypothesis on colonocyte migration within the mucocellular layer discussed earlier in this review, we have developed a simple sampling device for the direct collection of exfoliated cells from the surface of human rectal mucosa with minimal invasiveness.46 The author believes that discussing promising preliminary results of our ongoing trials here is not appropriate since they have only been reported in a meeting,47 but not published so far. At the same time presenting a few examples of cellular material collected from the surface of rectal mucosa (see Fig. 3) is essential for this review. One remarkable observation (Fig. 3f) clearly indicates that cells exfoliated from the surface of proximal colorectal cancers can reach rectum in the mucocellular layer without being lost. Although various aspects of the new technique require further investigation, there are good reasons to hope that direct collection of exfoliated cells from the surface of rectal mucosa can be preferable compared with stool-based techniques in terms of providing high quality material for a wide range of diagnostic and research applications.
Exfoliated colonocytes, DNA extracted from stool and detection of molecular biomarkers of colorectal cancer
It has already been mentioned that with the development of the Vogelstein's genetic model for colon carcinogenesis in the early nineties,1 the possibility of using tumour-derived DNA for colorectal cancer screening and early diagnosis attracted considerable attention. Initial report on the demonstration of the presence of K-ras oncogene mutations in DNA isolated from stool samples of patients with colorectal malignancies51 provoked a surge of interest to the detection of cancer-associated molecular markers in the human DNA present in faeces. Although it is usually presumed that this DNA originates exclusively from exfoliated colonocytes excreted with stool,21, 52, 53, 54, 55, 56, 57, 58 one can argue that well-preserved squamous epithelial cells (see Fig. 3d) mechanically sloughed in abundance from the surface of the anal canal epithelium during every stool passage become attached to the stool surface and can considerably contribute to the final DNA yield. The impact of this DNA, which certainly “dilutes” colonocyte-derived material, should not be disregarded as an interfering influence in detecting the molecular changes relevant for colorectal cancer.
The most important problem with the diagnostic value of stool DNA analysis is related to the apparent complexity of colorectal carcinogenesis involving multiple metabolic and regulatory pathways governed by interacting networks of genetic control.3 At the present level of our knowledge there is no identified reliable single cancer-specific molecular biomarker (i.e., some “unifying” marker always present in tumour rather than normal cells making neoplastic cells unambiguously detectable). The main consequence of this unfortunate problem is the necessity of using overlapping sets of multiple complementary molecular markers (multitarget DNA assay panels or MTAPs) present in malignant cells at relatively high frequencies.21, 52, 53, 54, 55, 56, 57, 58
Moreover, from the methodological point of view, human stool is a difficult material for DNA extraction: the abundance of bacteria and cytolytic substances in faeces can interfere with both colonocyte DNA stability and isolation procedures, often resulting in the coisolation of undesirable bacterial DNA and substances interfering with PCR. For these reasons the task of obtaining representative amounts of high quality human DNA from faecal material is always challenging.55, 56, 59, 60
Although the stool DNA-based approach has already brought some promising results,52, 53, 54, 55, 56, 57, 58 it is evident that its introduction as a common tool for colorectal cancer screening and early diagnosis is still impossible. Given that on one hand the only large prospective screening trial testing this approach resulted in a sensitivity of 51.6% (specificity 94.4%),57 and on the other hand the cost of the test comprising a multitarget panel of 21 molecular markers (detection of multiple point mutations in the sequences of the K-ras, APC and p53 genes, microsatellite instability of the Bat-26 gene and long fragment DNA analysis) is over $700 per assay (Pre Gen Plus test marketed by EXACT Sciences), we are still a long distance away from the ideal of cost-effective and simple colorectal cancer screening based on the molecular analysis of stool-derived DNA. Using a Markov model for cost-effectiveness assessment for mass colorectal cancer screening, Song et al.61 concluded that the approach can become marginally cost-effective only at $195 per assay presuming that its sensitivity reaches 65% for cancers and 40% for large polyps. Although rapid methodological progress allows room for future optimism, it is obvious that the assessment of stool samples for the presence of molecular cancer markers is excessively expensive in its present form.61, 62, 63
The use of relatively uncontaminated exfoliated colonocytes collected from rectum47 might help in overcoming problems associated with the analysis of stool samples, but additional research is requited to prove this point.
It should also be taken into account that recent advances in the understanding of the structure and function of the human genome have resulted in a breathtakingly rapid development of such complex disciplines as genomics and proteomics making possible targeted detailed investigation of disease-related gene expression patterns. Presently there are only a few reports describing assessment of specific RNA expression in human stool samples42, 64, 65; however, the opportunity of using directly collected colonocytes for RNA isolation should also be considered as a possible alternative to stool-based techniques.
The place of exfoliated colonocyte analysis among other methods applied for colorectal cancer screening and early diagnosis
The problem of colorectal cancer screening has already been thoroughly discussed in a number of recent publications52, 66, 67, 68; therefore, it is considered here very briefly, with the emphasis on the possible use of exfoliated colonocytes. Although benefits of the active screening are now generally accepted,52, 66, 67, 68 the disease remains among the leading causes of oncological mortality. The search for suitable approaches to colorectal cancer screening and early diagnosis is going on for decades, but the choice of available techniques is still strictly limited, faecal occult blood testing (FOBT) being the most common method.52, 69, 70 Advantages and drawbacks of possible screening applications of full colonoscopy and flexible sigmoidoscopy as well as new approaches such as virtual colonoscopy and detection of molecular biomarkers in faeces (see previous section) have been discussed in several recent reviews.21, 52, 71, 72 Unfortunately, neither of these methods can offer the desirable combination of low invasiveness, simplicity and affordable cost with high sensitivity and specificity. FOBT is noninvasive, cheap and simple, but all versions of this test produce high rates of both false-negative and false-positive results.21, 52, 70, 73, 74 Even its reported reducing effect on the incidence of colorectal cancer69 is now questioned.75 Colonoscopy, when performed by an experienced specialist, can be regarded as a precise and reliable diagnostic procedure; however, it is invasive and expensive,21, 52, 71 thus its use for routine population mass screening appears to be problematic. Moreover, the cumulative risk of complications becomes a sizable negative factor in repeated screening colonoscopies of large groups of predominantly disease-free people.21 Limitations of the flexible sigmoidoscopy are obvious since it allows only distal colon examination. The recently introduced computed tomographic colonography (CT colonography or virtual colonoscopy) is comparable with colonoscopy in terms of sensitivity and specificity72; however, high cost seriously hampers expectations of making this approach a mass screening method. In other words it appears that colonoscopy or its virtual analogue can be especially useful and cost-effective when applied as a “second line” or confirmatory screening in subjects preselected by a much wider “first line” screening procedure. The position of the latter is presently occupied by the inefficient FOBT.
Although rapid methodological progress in the development of molecular diagnostic procedures provokes optimistic predictions,21, 53, 71 the reality shows that most advanced molecular approaches based on the stool analysis are still too complicated and expensive to be seriously considered for mass screening (see previous section of this review). In this situation the possibility of getting well-preserved exfoliated colonocytes (see Fig. 3) directly from the surface of human colorectal mucosa may present an interesting option. The availability of this material can facilitate the application of both cell-based and molecular approaches to cancer biomarker detection. In addition, there may be a good opportunity of using cheap and straightforward quantitative methods such as the detection of the increased cell exfoliation from malignant tumours by simple DNA quantitation. We previously tried the latter approach using exfoliated colonocytes isolated from stool,25 but standardization difficulties made its validity questionable. The procedure of direct collection of exfoliated cells from rectum appears to be much easier to standardize.46 Our preliminary results on the use of this material for DNA quantitation look encouraging47; however, these findings need to be meticulously scrutinised and confirmed in clinical trials.
This review has briefly addressed problems surrounding the investigation of colonocyte exfoliation in the human colon. Although there was enough controversy and misunderstanding about the destiny of exfoliated colonocytes, it now appears that some basic principles of cell exfoliation in normal conditions and neoplasia finally emerge from obscurity. Malignant colorectal tumours definitely exfoliate huge amounts of cells, which mostly remain within the mucocellular layer overlying colorectal mucosa. Cell-containing fragments of the mucocellular layer excreted with stool are likely to constitute the main source of exfoliated colonocytes found in human faeces. Recent findings indicate that there may be a possibility of migration of the cell-containing mucocellular layer toward rectum creating conditions for the accumulation of exfoliated colonocytes on the surface of the rectal mucosa. It appears that direct collection of these cells can provide valuable material for various types of analysis. Methodological approaches valid for the development of new diagnostic and screening strategies for colorectal cancer based on the use of exfoliated colonocytes may include both advanced cytological techniques and the employment of cell-derived DNA, RNA and proteins for the identification of molecular biomarkers of neoplasia. Nevertheless, further profound research is required to address numerous unanswered questions still existing in the area of colonocyte population dynamics in normal conditions and cancer.
The author thanks the members of the Colonix Medical team (Dr Tatiana Bandaletova, Dr Andrew Llewelyn, Rupert Lywood, Per Aniansson, Hugo Lywood) for their kind support and to our clinical collaborators (Prof. Christopher Marks, Dr Jeremy Gibson, Dr Colin Ferrett) for their efforts in providing us with excellent clinical materials relevant for this work. The author thanks Dr Tatiana Bandaletova for her help in preparing and selecting microphotographs.