Chromosomal Alterations during Lymphatic and Liver Metastasis Formation of Colorectal Cancer

Abstract

Comparative genomic hybridization (CGH) was used to screen colorectal carcinomas for chromosomal aberrations that are associated with metastatic phenotype. In total, 63 tumor specimens from 40 patients were investigated, comprising 30 primary tumors, 22 systemic metastases (12 liver, 6 brain, and 4 abdominal wall metastases) and 11 lymph node tumors. Using statistical analysis and histograms to evaluate the chromosomal imbalances, overrepresentations were detected most frequently at 20q11.2–20q13.2, 7q11.1–7q12, 13q11.2–13q14, 16p12, 19p13, 9q34, and 19q13.1–19q13.2. Deletions were prominent at 18q12–18q23, 4q27–4q28, 4p14, 5q21, 1p21–1p22, 21q21, 6q16–6q21, 3p12, 8p22–8p23, 9p21, 11q22, and 14q13–14q21. Hematogenous metastases showed more alterations than lymph node tumors, particularly more deletions at 1p, 3, 4, 5q, 10q, 14, and 21q21 and gains at 1q, 7p, 12qter, 13, 16, and 22q. Comparing liver metastases with their corresponding primary tumors, particularly deletions at 2q, 5q, 8p, 9p, 10q, and 21q21 and gains at 1q, 11, 12qter, 17q12–q21, 19, and 22q were more often observed. The analysis suggested that the different pathways of tumor dissemination are reflected by a nonrandom accumulation of chromosomal alterations with specific changes being responsible for the different characteristics of the metastatic phenotype.

Introduction

Cancer of the colon and rectum is the second most prevalent cause of cancer deaths in men and the third most common in women [1]. Metastasis is responsible for most cancer deaths. It represents the essential step during cancer progression from a locally growing tumor to a metastatic killer. This switch is believed to involve numerous alterations that allow tumor cells to complete the complex series of events needed for metastasis. Relatively few genes have been implicated in these events. Therapeutic strategies to prevent the development of metastases thus have potential impact on cancer mortality. The development of these therapies requires a better understanding of the biology and molecular events of the metastatic process. Therefore, a primary goal in our research is the understanding of the molecular mechanism mediating not only the development of primary colorectal cancers but also the process of colon carcinoma metastasis.

Metastasis is defined as the spread of cells from a primary neoplasm to distant secondary sites and proliferation at these sites. This highly selective process consists of a series of linked, sequential steps favoring the survival of a subpopulation of metastatic cells that preexist within the primary tumor mass [2]. These steps include the detachment of cells from the primary tumor, invasion of the surrounding tissues, penetration into the circulation (bloodstream or lymphatic system), implantation into the capillary beds of target organs, extravasation and invasion into the target tissue, formation of vascular network, and, finally, proliferation at this secondary site of implantation [2,3]. For production of clinically or microscopically detectable metastases, each of these steps must be completed. Failure to complete even one step in this process (e.g., the inability to invade host stroma, a high degree of antigenicity, inability to grow in a distant organ's parenchyma) results in tumor cell elimination or dormancy. The successful metastatic cell must, therefore, exhibit a complex phenotype that experimental evidence suggests to be regulated by transient or permanent changes in different genes at the DNA and mRNA level(s) [3]. Numerous examples exist in which malignant tumors metastasize to specific organs. Paget [4] proposed in 1889 that the organ microenvironment (the soil) can influence the implantation, invasion, survival, and growth of particular tumor cells (the seeds). This hypothesis explains colonization patterns that cannot be explained solely by mechanical lodgment theories and anatomical considerations [5].

Recently, we hypothesized that distinct chromosomal alterations of the primary tumors may predispose to dissemination and may thus play a role in the early phase of metastasis formation (metastasis initializer lesions), whereas others are responsible for the peculiarities of the metastatic phenotype-like organ-specific metastasis formation and may be preferentially acquired during later phases (metastasis modulator lesions) [6]. It is interesting to note that many metastasis-associated lesions are detectable in the primary tumor, questioning the model of a stepwise cancer progression [7].

To provide cytogenetic data, we investigated 63 metastatic colorectal carcinomas, 30 primary carcinomas, and 33 metastases from different sites by comparative genomic hybridization. Using difference histograms and case-by-case histograms for the analysis of paired tumor samples from the same patient, we were able to identify recurrent novel chromosomal changes in different metastatic sites.

Materials and Methods

Patients and Tumor Samples

The study consisted of 63 specimens from 40 patients, which were mainly obtained during surgical resections at the Department of Surgery of the Charité at Campus Buch. Additionally, tumor specimens of the brain were collected during neurosurgical operations at the Charité Hospital. The tumor collective and its clinicopathological data according to the TNM criteria of the UICC are summarized in Table 1. Nineteen of 35 patients showed hematogenous and lymph node metastases. Liver and nodal metastases were observed in nine patients. In all brain metastases, the nodal status was unknown and the four abdominal wall metastases showed lymph node metastases in two patients, no lymph node metastases in one patient, and unknown nodal status in the other patient.

Table 1.

Tumor Collective.

Number of specimens	63
Number of patients	40
Primary tumors	30
Metastases	33
Liver	12
Lymph nodes	11
Brain	6
Abdominal wall	4
Dukes stage of patients
B	3
C	2
D*	35

^*Nineteen of 35 patients with hematogenous spread (pM1) showed additional lymph node metastases (5 pN0, 5 pN1, 14 pN2, 11 pNX; X = status unknown).

One aliquot of tumor tissues was frozen in liquid nitrogen and kept at -80°C until DNA extraction. DNA was extracted from several 30-µm cryostat tissue sections by proteinase K digestion and phenol chloroform extraction, which was verified to consist of a minimum of 70% tumor cells in each case. A second aliquot was submitted to formalin fixation and paraffin embedding.

Comparative Genomic Hybridization (CGH)

DNA labelling, hybridization, and detection were performed as previously described [6]. The protocols are also available at our web site (http://amba.charite.de/cgh). DNA was extracted from several 30-µm cryostat tissue sections by proteinase K and phenol chloroform extraction, which was verified to consist of a minimum of 70% tumor cells in each case.

Digital Image Analysis

Image acquisition and digital image analyses have also been described in detail [8]. At least 15 metaphases/karyograms were analyzed per case, calculating CGH sum karyograms and mean ratio profiles with confidence intervals. Briefly, the deviations of the mean FITC/TRITC profiles from the normal ratio of 1.0 were tested for significance by a two-sided Student's t-test. Deviations of the ratio profile with at least 99% significance in the Student's t-test were scored as DNA gains or losses (i.e., only those imbalances in which the ratio profile with its 99% confidence interval exceeded the line of the normal ratio 1.0 to the same side were included in the evaluation). This procedure is rather sensitive for scoring chromosomal alterations by CGH [6,8]. Pronounced DNA gains and losses of the tumor collective shown in the histogram of Figure 1A were defined by those alterations exceeding the ratio values 1.5 and 0.5, respectively. They most likely correspond to high copy amplifications or multicopy deletions.

Figure 1.

(A) Summary of all genetic alterations of 63 advanced colorectal carcinomas in a histogram representation. The chromosomal imbalances are shown as incidence curves along each chromosome. Areas on the left side of the chromosome ideogram correspond to loss of genetic material; those on the right side correspond to DNA gains. The changes were determined by a statistical method. Those with 99% significance in the Student's t-test are shown in blue; the additional ones with only 95% significance are depicted in green. Pronounced DNA gains and losses, defined as regions where the ratio profile exceeded the fixed thresholds 1.5 and 0.5, are shown in red and most likely correspond to high copy amplification and multicopy deletions, respectively. Heterochromatic areas (centromeric and paracentromeric regions of chromosomes 1, 9, and 16; p arms of acrocentric chromosomes) must be excluded from the analysis. (B) Difference histogram of lymph node metastases (n = 11) versus liver metastases (n = 12). The red areas represent the percentage of changes that are present only in liver metastases, whereas the green color indicates the excess of changes of lymph node metastases. The white areas beneath the colored part of the histogram represent the percentage of changes that are present in both tumor groups. The liver metastases (hematogenous) showed more alterations than lymph nodes tumors (lymphogenous) with significant changes as indicated by the gray horizontal lines (light gray, regions with 95%; dark gray, regions with 99% significance; chi-square test) (e.g., overrepresentation at 1q22–q23 and deletions at 4p). (C) Case-by-case histogram of 10 paired samples (i.e., primary tumor and corresponding liver metastasis). Four conditions are represented: blue, percentage of chromosomal imbalances that are common in both tumors; green: those chromosomal imbalances that are additionally seen in the primary tumor; red, those that are extra present in metastases; yellow, proportion with no changes in both groups. The blue areas are dominating again, reflecting the high concordance between primary tumors and metastases. However, the metastases showed additional alterations at several regions (e.g., gains at 1q and losses at 2q, 5q31, 8p, and 21q).

Comparison of Tumor Subgroups (e.g., Lymph Node Tumors and Liver Metastases)

A difference histogram (Figure 1B) was generated as described to compare the lymph node tumors and liver metastases as tumor subgroups [6]. Only alterations with 99% significance were included in this analysis. The percentage of changes occurring only in liver metastases is represented by the red color, whereas the excess of changes in the lymph node tumors is shown in green. The white areas beneath the colored part of the histogram represent the percentage of changes that are present in both subgroups. A large colored area thus indicates a pronounced difference between the tumor groups. The differences were tested for significance by a chi-square test. Areas with 95% significance (.01 < P < .05) are depicted in bright grey, and areas with 99% significance (P < .01) are depicted in dark grey horizontal lines.

In addition, we applied our newly developed case-by-case histogram in this study. It visualizes the comparison of paired samples (i.e., the primary tumor and its corresponding metastasis). Within the histogram, the changes are differentiated into four types [i.e., imbalances that are common in both tumors (blue color), those that are present additionally in the primary tumor (green), those that are present additionally in the metastasis (red), as well as no changes at all in both groups] (Figure 1C, yellow).

Results

Histogram Analysis of All Colorectal Carcinomas and Specific Subgroups

The chromosomal imbalances of the 63 colorectal carcinomas were first summarized by a histogram (Figure 1A). It represents the DNA gains and losses as an incidence curve along each chromosome. Additional histograms were calculated for the subgroups of primary tumors, lymph nodes, and liver metastases. For the comparison of these groups, difference histograms (Figure 1B) as well as case-by-case histograms (Figure 1C) were generated, which are exemplified in Figure 1.

In general, regional (lymphogenous) metastases carried less copy number aberrations than hematogenous metastasis (Figure 1B). Only a few changes were more frequently found in the lymph node tumors compared to the primary tumors (i.e., deletions of chromosome 18 and 21q). In contrast, many more alterations were seen in hematogenous metastases (i.e., deletions at 1p, 3, 4, 5q, 10q, 14, and 21q and gains at 1q, 7p, 12qter, 13, 16, and 22q). Comparing liver metastases with their corresponding primary tumors, particularly deletions at 2q, 5q, 8p, 9p, 10q, 11p, and 21q and gains at 1q, 11q, 12qter, 17q12–21, 19, and 22q were more often observed. The compilation of data is presented in Figure 2.

Figure 2.

Colorectal cancer progression model with typical morphological and chromosomal changes. Typical chromosomal imbalances associated with lymph node and liver metastasis formation are represented, DNA gains are shown in red; DNA losses in green. The two arrows indicate that lymphatic and hematogenous tumor spread may occur independently of each other, although there are alterations common for both dissemination pathways (e.g., deletions on 8p and 21q21). Additionally, there are characteristic morphologic features. Liver metastases typically show a kribriform solid tumor growth with pronounced apoptosis and necrosis with only little stroma, a pattern frequently detectable already in the primary tumor corresponding to a poorly differentiated carcinoma (G3, high-grade). In contrast, lymph node metastasis and nonmetastasizing primary tumors often show a predominant tubular differentiation with a strong desmoplastic stroma reaction and are thus classified as low-grade carcinomas (G1 and G2).

Discussion

This study is the first comprehensive and largest analysis of DNA gains and losses associated with lymphatic and liver metastasis formation of colorectal cancer. Investigating metastasizing colorectal carcinoma resection specimens, we were able to study 30 primary tumors and 33 metastases by CGH. As a result, recurrent chromosomal regions involved in the neoplastic progression of colorectal cancer and their different dissemination pathways were detected. The fact that we confirmed several changes that have been previously reported using other methodologies supports the validity of our findings. Furthermore, we analyzed tumor pairs (i.e., primary versus metastatic carcinoma or lymph node versus liver metastasis) from the same patient, which is, in our opinion, an appropriate study design to identify genetic alterations in metastasis.

The CGH data, together with typical morphological figures, were represented in the progression model of Figure 2. It is important to note that there are often characteristic morphologic differences between tumors at distinct sites. Liver metastases typically show a kribriform solid tumor growth with pronounced apoptosis and necrosis with only little stroma, a pattern being frequently detectable already in the primary tumor corresponding to a poorly differentiated carcinoma (G3, high-grade). In contrast, lymph node metastases and nonmetastasizing primary tumors often show a predominant tubular differentiation with a strong desmoplastic stroma reaction and are thus classified as low-grade carcinomas (G1 and G2). These morphologic differences are probably associated with the distinct biologic behavior. Additionally, they may induce a technical bias (i.e., a more prominent normal cell contamination in lymph node metastasis and primary tumors compared to hematogenous metastasis in parenchymal organs such as the liver). We checked our fresh-frozen tumor samples in each specimen by hematoxylin and eosin (HE) staining and used macrodissection to minimize the amount of contaminating normal tissue. Although this may still influence the analysis, we are convinced that the higher overall number of genomic alterations in the liver metastases cannot be explained by these methodological considerations. This is a confirmation of our previous results indicating that additional alterations occur during hematogenous dissemination, suggesting specific alterations being responsible for dissemination to the brain [9].

Many of the specific chromosomal imbalances that we identified (i.e., losses at 1p, 3p, 4, 5q, 10q, 14q, and 21q and gains at 1q, 11, 12qter, 17q12–q21, 19, and 22q) have already been associated with tumor progression and dissemination in colon cancer [6,10–12] or other cancer types [9,13–21]. Similarly, many candidate genes were already described (e.g., S100A4 and Cox-2 on 1q [22,23], BLU on 3p [24], APC at 5q21 [25], c-erb-B2 at 17q21 [26], AIB1 at 20q [14], and MMP11 at 22q12.2 [19]). In general, it is important to note that the chromosome losses and gains may not only be associated with the inactivation or activation of any specific tumor suppressor gene or protooncogene, but correspond to changes in the expression level of multiple genes, which may directly or indirectly affect tumor growth and dissemination. Furthermore, the chromosome regions may harbor several classical tumor suppressor genes (e.g., p14, p15, and p16 on chromosome 9p) [27,28].

Our data and the thereof derived model indicate that there are independent pathways of colorectal tumor dissemination and that these are associated with a nonrandom accumulation of chromosomal alterations underlying the different characteristics of the metastatic phenotype. It also highlights two well-known pathogenetic mechanisms: 1) metastasis formation may occur immediately after invasion of the primary tumor; and 2) hematogenous dissemination may occur independently from lymphatic tumor spread. Obviously, there is an overlap between the genetic alterations of the tumor subgroups correlating with the fact that many tumors develop both lymph node or hematogenous filiae. Lymphatic spread, however, is not a prerequisite of systemic dissemination, which may occur early after cancer initiation. Accordingly, there is probably no stepwise acquisition, but a selection of tumor cell clones carrying the favorable metastasis-associated lesions that are initially generated randomly by the inherent chromosomal instability of most colorectal carcinomas. In general, our model fits well with the alterations described previously [6,9,12,29–33]. Similar to the progression model of Fearon and Vogelstein [25] concentrating on genetic lesions, we would argue that the accumulation of specific chromosomal imbalances, rather than their order, is important for the progression of colorectal cancer.

In conclusion, the CGH data presented here define a set of novel genomic regions in the human genome that are likely to harbor genes that play an important role in the genesis of organ-specific metastasis. Further molecular genetic studies will identify more precisely these regions and will hopefully lead to the identification of relevant genes underlying the metastatic process.

Abbreviation

CGH

comparative genomic hybridization