| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
1 Experimental Research Unit, General University Hospital of Albacete, C/ Hermanos Falcó 37, 02006 Albacete, Spain
2 Department of Pathology, General University Hospital of Albacete, Albacete, Spain
3 Department of Surgery, Universidad Autónoma de Madrid and La Paz University Hospital, Paseo de la Castellana 261, 28046, Madrid, Spain
Correspondence: Address correspondence and reprint requests to: Dolores C. García-Olmo, PhD; E-mail: doloresg{at}sescam.jccm.es.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: We injected DHD/K12-PROb cells subcutaneously into BD-IX rats. Three weeks later, we removed the tumors surgically. Group PERI was sacrificed 3 hours after surgery, group POST-2 was sacrificed 2 weeks later, group POST-4 was sacrificed another 2 weeks later, and group POST-LONG was sacrificed when rats were close to death. In group PERI, four perioperative blood samples were taken. In the other groups, only one blood sample was taken per rat, immediately before euthanasia. We used polymerase chain reaction to detect tumor-derived DNA in buffy-coat, plasma, and lung samples.
Results: In group PERI, tumor DNA in plasma was more frequent than circulating tumor cells at all perioperative time points. The difference was statistically significant 3 hours after surgery (P = .035). In group POST-2, there was no detectable metastasis or tumor DNA in blood samples. There were lymphatic and lung metastases in most animals in group POST-4 and in all animals in group POST-LONG. In the last two groups, the frequencies of tumor DNA in the buffy coat and in plasma were similar.
Conclusions: In our animal model, the hematogenous dissemination of tumors due to surgery seemed to be more closely related to tumor-derived cell-free DNA than to circulating tumor cells. In addition, the surgical resection of primary tumors did not inhibit the development of metastases.
Key Words: Circulating tumor cells • Cell-free tumor DNA • Colon cancer • Surgery • Metastasis
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although these effects of surgery are initiated simultaneously and might act synergistically, some authors have emphasized the role of immunosuppression in the seeding of metastases at distant sites.1,2 Others have suggested a clear relationship between this shedding and the active production of oxygen by leukocytes that is due to excessive surgical stress.3 Many other researchers have focused their studies on the effects of the surgical manipulation of tumors on the hematogenous dissemination of cancer cells.
The surgical manipulation of tumors can be associated with the shedding of tumor cells into the bloodstream, and such shedding has been considered essential for the hematogenous metastasis of solid tumors.4–6 However, despite the large number of studies of circulating tumor cells during the perioperative period,4–10 this issue remains a focus of debate because reported results and their interpretations have been inconsistent and, sometimes, contradictory. It is possible that the heterogeneity of results might be due to the nature of the cancer studied and its location, to the surgical techniques used, to the techniques used to detect circulating tumor cells, or to differences in the sizes of the populations studied.
Taking a different approach, many authors have suggested the possible utility of the analysis of cell-free tumor-derived nucleic acids in plasma as a valuable diagnostic11,12 and prognostic13,14 tool. Some studies have focused on the postoperative detection of tumor DNA in plasma after surgical treatment of a variety of cancers, and the results suggest the possible prognostic value of the presence of plasma DNA after surgery.15–17 However, to our knowledge, no study has reported the presence of tumor DNA in the plasma during the perioperative period.
The presence of cell-free DNA in cancer patients has been assumed, frequently, to be a simple consequence of tumor growth rather than a phenomenon that might be involved in tumor progression. However, a possible role has been suggested for cell-free tumor nucleic acids in the progression of tumors and the development of metastases by genetic horizontal transfer or related mechanisms.18,19 Indeed, this hypothesis of genometastasis might be supported by a variety of findings (see review19). It has been proposed that cell-free nucleic acids not only might be a product of the degradation of tumor cells, but also might be released from cells by an active mechanism.20–22 In addition, it has been shown repeatedly that nucleic acids in serum have sufficient integrity to allow amplification and that such integrity might be due to a protective effect of vesicles or membranes (see review19). Thus, we postulated that it might be interesting to examine the effects of a surgical procedure on the release of tumor DNA into the bloodstream and the possible implications of such release in metastasis.
In a previous study, we used a heterotopic model of colon cancer in the rat and found that cell-free tumor DNA was detectable in plasma sooner and more frequently than were circulating tumor cells during progression of the tumor.23 Our results suggested that tumor DNA in the plasma might be much more common than detectable hematogenic tumor cells during the spread of colorectal cancer.23
In this study, we used the same animal model of cancer to compare the rates of detection of tumor-derived DNA in the buffy coat and in plasma from tumor-bearing rats before and after surgery. We attempted to correlate these rates with surgical manipulation and the outcome of surgery.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Animals
Both male and female BD-IX rats were used. They were taken from a colony established at the authors’ animal facility from founders purchased from a commercial breeder (Charles River Laboratories, Barcelona, Spain). Breeding was performed in compliance with European Community Directive 86/609/ CEE for the use of laboratory animals. As recommended by the Federation of European Laboratory Animal Science Associations, rats in the animal facility are tested periodically to ensure that the colony is free of pathogens such as Mycoplasma pulmonis, Salmonella sp., Sendai virus, Hantaan virus, and Toolan H1 virus.
From birth to the end of the experiments, all rats had unlimited access to water and standard rat chow (Panlab s.l., Barcelona, Spain). At the beginning of the experiments, rats were 6 to 8 weeks old and weighed 95 to 200 g.
Implantation of Tumors
Tumors were generated in the thoracic region by unilateral subcutaneous injection of DHD cells into the right side of the chest. Cells were trypsinized, washed, and resuspended in phosphate-buffered saline. Then, .25 mL of this suspension, containing 1 x 106 cells, was injected per rat.
The growth of subcutaneous tumors was monitored in all animals and recorded weekly. We measured the greatest diameter of each tumor with electronic calipers.
Surgical Resection of Tumors and Design of Experiments
Three weeks after the injection of tumor cells, tumors were surgically removed. Before surgery, rats were anesthetized with an intraperitoneal injection of a mixture of ketamine (75 mg/kg) and xylazine (10 mg/kg). Then we made an incision in the skin over the thoracic wall, circling the tumor mass with a margin of approximately 4 mm. Tumors were not excised from the skin but were removed with it (Fig. 1
). In this way, we always were sure that no tumor mass remained attached to the skin. The size of tumors at the moment of surgery and the resilience of the skin of the rats allowed us to perform this maneuver and, subsequently, to suture the skin without tissue tension. The incision was closed with four or five stitches with 3-0 silk suture.
|
Collection of Blood and Tissue Samples
In the case of animals in group PERI, four blood samples were taken in the perioperative period: before incision, after removal of the tumor, immediately after skin suturing, and 3 hours later. All blood samples were extracted by cardiac puncture with the animals under anesthesia. The volume of each of the first three samples was 1.0 mL, and that of the last was 2.0 to 4.0 mL. After the last sample had been collected, a lethal dose of sodium thiopental was administered intracardially. In the case of animals in the three other groups, only one blood sample (approximately 4.0–4.5 mL per rat) was taken, and it was taken by cardiac puncture immediately before euthanasia.
All samples of blood were collected in tubes with ethylenediaminetetraacetic acid and subjected to density gradient centrifugation through Ficoll (Ficoll-Paque Plus; Amersham Pharmacia, Uppsala, Sweden) for isolation of the buffy coat and plasma.
Lungs were inspected visually for the presence or absence of pulmonary metastases, which was recorded. When no macrometastasis was found, samples of lungs were removed, frozen, and stored at –80°C before extraction of DNA and amplification by polymerase chain reaction (PCR) for the detection of tumor DNA.
Although we reported previously that, in this animal model, only lung metastases develop,23 we also recorded the presence or absence of peritoneal met-astatic dissemination and other affected organs. To this end, we made two perpendicular incisions in the abdomen of each rat to allow a comprehensive inspection of the cavity on separation of the edges of the incision.
Detection of Tumor DNA in Blood and Lung Samples by PCR
We performed mutant allele–specific amplification by PCR to examine the presence of a point mutation (GGT 
GAT) in the k-ras oncogene (exon 1; codon 12) that is found in DHD cells. To ensure the validity of the technique, all samples of DNA were also amplified to examine the presence of nonmutated k-ras under the same conditions. The sequences of the forward primers were 5'-CTTGTGGTAGTTGGA GCTGA-3' for detection of mutated DNA and 5'-CTTGTGGTAGTTGGAGCTGG-3' for detection of wild-type DNA. In both reactions, the reverse primer was 5'-GCCACCCTTTACAAATTGTA-CAT-3'. The predicted length of the amplified fragment of interest was 140 base pairs.
DNA was extracted from the samples of buffy coat, plasma, and lung with three commercial kits, namely, the High Pure PCR Template Preparation Kit (Roche Diagnostics GmbH, Mannheim, Germany), QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany), and QIAamp DNA Mini Kit (Qiagen), respectively.
The reaction mixture for PCR, with a total volume of 50 µL, contained a minimum of 10 ng of template DNA in a volume of 5 µL for the samples of buffy coat and lung and in a volume of 7 µL for samples of plasma. To avoid unspecific reactions (false-positive results), the amount of DNA did not exceed 250 ng. Thus, it was necessary to dilute the DNA collected from many samples of the buffy coat and lung before PCR. The reaction mixture also contained .8 µM of each primer, .2 mM of deoxy-nucleosides triphosphate solution, 1.5 mM of MgCl2, 5 µL of 10 x buffer, and .4 U of Taq DNA polymerase (Roche Diagnostics GmbH) for the samples of buffy coat and lung and 1 U for the plasma. Mutant allele–specific amplification PCR was performed with an initial sample-denaturation step of incubation at 94°C for 10 minutes. Then amplification was allowed to proceed for 35 cycles of incubation for 1 minute at 94°C, for 30 seconds at 63°C, and for 45 seconds at 72°C, with final extension for 10 minutes at 72°C. Temperature cycling was achieved with a DNA thermal cycler (iCycler; Bio-Rad, Hercules, CA).
The products of amplification were subjected to electrophoresis on a 3% agarose gel in tris-borate-ethylenediamine tetraacetic acid (Tris-borate-EDTA) BE (Pronadisa, Madrid, Spain) that contained 2 µg/ mL of ethidium bromide. Bands of DNA were visualized under UV light and photographed on Polaroid film (no. 667; Polaroid Co., Cambridge, MA).
For each PCR, three controls were included to detect DNA contamination, false-negative results, and false-positive results. For these controls, we used, respectively, water (no added DNA), DNA from DHD cells, and DNA from the buffy coat, plasma, or lungs of healthy animals, as appropriate (Fig. 2).
|
2 test and Fisher’s exact test when any frequency was <5. Two-tailed P values <.05 were considered statistically significant. Statistical analyses were performed with EPI-INFO, version 6.04 (Centers for Disease Control and Prevention, Atlanta, GA).
Ethical Considerations
All provisions related to the care and handling of animals used for experiments and required by Spanish law and specifications of the European Community were applied.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Tumor Recurrence and Metastases
No tumor was found at the site of the incision in any animal during the postoperative period. There were no tumors in the right axillary lymph nodes of animals in group POST-2, but they were found in three of the five animals in group POST-4 and in all those in group POST-LONG (Table 1). In group POST-4, tumors were palpated 3 weeks after surgery, and they continued to grow until the rats were sacrificed, a week later. Because the right axillary lymph node was close to the site of the primary tumor, lymphatic metastases might have been confused with tumor recurrence. To avoid such confusion, we verified the lymphatic nature of postoperative tumors by histological staining and microscopic observations of tumor samples after necropsy. In all cases, lymph tissue could be identified (Fig. 3).
|
|
). In one animal, such tumors were first palpated at 21 weeks (Fig. 4). In all cases, tumors were initially palpated in the axillary area, but they grew and descended to occupy a large area of the right thoracic wall. The mean tumor diameter at sacrifice was 2.2 ± .6 cm. Lymph tissue was not detected in the histopathological study of samples from this group because tumor cells had invaded the entire nodes.
|
). The relationship between the presence of lymph and lung metastases was statistically significant (P < .001).
As mentioned in Materials and Methods, we analyzed lung samples by PCR only when no macrometastases were found. No tumor DNA was detected in any of these samples.
Survival of Animals
The survival of animals in group POST-LONG ranged from 59 to 151 days after surgery (mean ± SD, 90.0 ± 35.8 days; Fig. 4).
Detection of Circulating Cancer Cells
All samples of buffy coat yielded the wild-type (nonmutated) k-ras sequence. Detection of tumor DNA (the mutated k-ras sequence) in samples of buffy coat was considered to be an indicator of the presence of circulating tumor cells.
In the animals in group PERI, tumor DNA was found in two of the samples of buffy coat taken before surgery, in two samples taken during surgery, in one sample taken after surgery, immediately after closing, and in none of the samples taken 3 hours later (Table 2). All the positive samples came from only two animals. In other words, circulating tumor cells were detected in the blood samples from only 2 of 11 animals, and, in both cases, preoperative samples were positive (Table 2).
|
). The difference between groups was not statistically significant. Circulating tumor cells were detected in only 3 of the 10 animals with lung macrometastases and in 3 of the 10 animals with lymphatic metastasis. The detection of circulating tumor cells was not related, statistically, to the presence of macrometastases.
Detection of Tumor DNA in Plasma
All samples of plasma yielded the wild-type k-ras sequence (Fig. 2). Detection of tumor DNA in plasma samples was considered to be an indicator of the presence of cell-free tumor-derived DNA.
In group PERI, the four plasma samples from one animal were contaminated by exogenous DNA during the DNA-isolation process and were excluded from the study. Thus, we could analyze the samples from only 10 tumor-bearing rats. We detected tumor DNA in at least one of the plasma samples from each of five animals (50%). In two of them, tumor DNA was not found in the preoperative samples of plasma and was found only in plasma obtained during or after surgery (Table 2).
Detection of tumor DNA in the plasma was more frequent at all times during the perioperative period than detection of circulating cancer cells. However, the difference was statistically significant only for samples obtained 3 hours after surgery (P = .035).
Moreover, no tumor DNA was detected in plasma from animals in group POST-2, but it was detected in two animals (40%) in group POST-4 and in one animal (14%) in group POST-LONG (Table 1).
The difference between groups was not statistically significant, and the detection of tumor DNA in plasma was not related to the presence of lymph node or lung macrometastases. There was no statistically significant difference between the frequencies of detection of tumor DNA in plasma and in samples of buffy coat during the postoperative period. Furthermore, there was no statistically significant correlation between the detection of tumor DNA in the plasma and in the buffy coat.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Guller et al.9 suggested that detection of tumor cells in the blood and in peritoneal lavage specimens might have potential clinical utility as a prognostic marker. However, their results contradicted the hypothesis that surgical manipulation enhances the dissemination of tumor cells during curative resection of colorectal cancer.9 In fact, their surgical manipulations were unrelated to any increase in the rate of detection of tumor cells in their samples of blood or even in their samples of peritoneal lavage fluid.
Other authors who have suggested that surgical manipulation might enhance the release of tumor cells into the circulation supported this suggestion with the results of an analysis of patients whose peripheral blood samples, obtained before and/or after surgery, gave PCR-positive results.5 However, most patients in the cited study had entirely negative samples of peripheral blood, as was the case in some other studies that focused on the perioperative period of the surgical treatment of solid tumors.8–10,24 Moreover, in most studies reported in the literature and cited here, patients in whom circulating tumor cells were detected only during and/or after surgery were in the minority,4,5,7–10,24–26 no matter what the conclusions of the authors. In our present study, most of the animals had PCR-negative perioperative samples of buffy coat, and in no animal were circulating tumor cells detected only during and/or after surgery. These observations contradict the hypothesis that surgical manipulation provokes the release of tumor cells.
It has been suggested that the incidence of tumor cell shedding is related to the degree of intraoperative manipulation. Thus, for example, patients undergoing major liver resection for metastases of colorectal cancer might be at higher risk for the intraoperative dissemination of tumor cells than patients undergoing resection of primary colorectal cancer or minor liver resection.7 However, this issue also remains debatable. In fact, Sheen et al.10 analyzed a group of patients with hepatocellular carcinoma, most of whom underwent major liver resection, and they did not find a statistically significant increase in the rate of detection of circulating tumor cells in the perioperative and postoperative period.
In our study, the surgical procedure was not aggressive because tumors were subcutaneous and were removed without excision from the skin. Thus, theoretically, the likelihood of dissemination of tumor cells as a result of mechanical manipulations should be low. We did, indeed, find a low rate of detection of perioperative circulating tumor cells. However, we found lymphatic and/or lung metastases in most animals sacrificed 4 weeks after surgery or later. Thus, although the low rate of perioperative detection of circulating tumor cells might have been due to the limited aggressiveness of the surgical technique, this low rate was independent of the development of metastasis. Our results do not support the proposal of an essential role for circulating tumor cells in the development of metastasis after surgery.
The rates of detection of tumor DNA in plasma were, however, higher than those of circulating tumor cells at all times during the perioperative period. The difference was statistically significant at the latest time analyzed, namely, 3 hours after surgery (P = .035). This result suggests that hematogenous dissemination by surgery might be more closely related to cell-free DNA than to circulating tumor cells. In a recent study, Yu et al.27 analyzed the effect of a percutaneous liver biopsy on the dissemination of tumor DNA into the plasma and found no evidence that such biopsies provoked the release of tumor DNA. However, to our knowledge, the present study is the first to examine the release of tumor-derived cell-free DNA in relation to the surgical resection of tumors. In light of our results, if we accept the possibility that cell-free DNA derived from tumors might have transforming ability, as proposed by the hypothesis of genometastasis,18,19 then it seems that the no-touch isolation technique might be preferable during tumor resection.
With respect to the groups of animals designed to evaluate the outcome after surgery (groups POST), we compared the results with those of a previous study of untreated animals, in which we had used the same cancer model.23 We found that the incidence of lung macrometastases 5 weeks after inoculation was 31% in untreated animals, whereas it was 0% in surgically treated animals. However, 2 weeks later, the frequency metastasis was slightly higher for treated (60%) than for untreated (45%) animals. It seems that, in this animal model, in which cancer is aggressive and metastatic, surgical treatment delayed but did not inhibit the development of metastases.
In our untreated tumor-bearing rats, we were able to detect lung micrometastases by PCR in many animals that did not have macrometastases.23 Specifically, at the end of the fifth and seventh weeks after inoculation, the frequencies of animals with lung micrometastases were 44% and 67%, respectively.23 However, in this study, we did not detect any micrometastases by the same PCR-amplification method in any of the treated animal that did not have lung macrometastases and that were sacrificed at the same times. In other words, after tumor resection, we could not detect lung metastases at early stages. This result suggests that, after a period of apparent cure, lung metastases developed more quickly than in untreated animals, a suggestion that would be in agreement with the hypothesis that tumor surgery might itself promote the development of metastases. However, there is another possible explanation for this finding. After surgery, metastases were macroscopic because they were seeded before surgery (at the first stages of tumor development), and micrometastases could not be detected because there was no new seeding after surgery. In addition, the mean duration of survival of animals in group POST-LONG was not significantly longer than that reported for untreated animals in the same model of cancer.28 In that previous study, in which the survival was recorded from the inoculation of tumor cells to the natural death, the mean duration of survival was 94.5 days.28 In this study, this mean was 111.0 days; however, the variation among animals was high (SD, 35.8 days), and this apparent increase of survival was due to the higher survival of only one animal (Fig. 4).
We detected neither circulating tumor cells nor tumor DNA in plasma from animals sacrificed at the end of the second week after surgery (fifth week after inoculation), which had neither macrometastases nor micrometastases in their lungs, as mentioned previously. It is of interest to compare these results with those obtained with untreated tumor-bearing rats because, in these animals, it was during the fifth week after inoculation that we obtained the highest rates of detection of circulating tumor cells and tumor DNA in plasma.23 Thus, it seems that surgical resection of the tumor provoked the clearance of circulating tumor cells and of tumor DNA in the plasma, which was related to an apparent cure. Some authors have reported the clearance of tumor DNA in plasma15,16 or of circulating tumor cells25 after tumor surgery, and the prognostic value of these phenomena is being discussed.
We did not find any statistically significant relationship between the detection of circulating tumor cells and tumor DNA in plasma and the presence of metastases. This result is in agreement with the results of our previous study of untreated tumor-bearing rats23 and with the reported absence of significant differences in the rates of detection of these cells and DNA between metastatic and non-metastatic stages in humans.13,29 However, this issue remains controversial.
In conclusion, in an animal model of metastatic colon cancer, the hematogenous dissemination of tumors due to surgery seemed to be more closely related to tumor-derived cell-free DNA than to circulating tumor cells. In addition, the surgical resection of primary tumors did not inhibit the development of metastases. By contrast, after a period of apparent cure, metastases seemed to develop more quickly than in untreated animals.
|
ACKNOWLEDGMENTS |
|---|
Received for publication May 31, 2005. Accepted for publication March 1, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
