Evaluation of 1p Losses in Primary Carcinomas, Local Recurrences and Peripheral Metastases from Colorectal Cancer Patients

Abstract

Cytogenetic and molecular genetic analyses of colorectal adenomas and carcinomas have shown that loss of the distal part of chromosome arm 1p is common, particularly in tumors of the left colon. Because the importance of 1p loss in colorectal cancer metastases is unknown, we compared the frequency, exact site and extent of 1p deletions in primary carcinomas (n=28), local recurrences (n=19) and metastases (n=33) from 67 colorectal cancer patients using 14 markers in an allelic imbalance study. Loss of 1p was found in 50% of the primary carcinomas, 33% of the local recurrences, and 64% of the metastases, revealing a significant difference between the local recurrences and the metastases (P=.04). The smallest region of 1p deletion overlap (SRO) defined separately for each group of lesions had the region between markers D1S2647 and D1S2644, at 1p35–36, in common. The genes PLA2G2A (1p35.1–36) and TP73 (1p36.3) were shown to lie outside this consistently lost region, suggesting that neither of them are targets for the 1p loss. In the second part of the study, microdissected primary carcinomas and distant metastases from the same colorectal cancer patients (n=18) were analyzed, and the same 1p genotype was found in the majority of patients (12/18, 67%). The finding that primary carcinoma cells with metastatic ability usually contain 1p deletions, and that some cases lacking 1p alterations in the primary tumor acquire such changes during growth of a metastatic lesion, supports the notion that 1p loss may be important both early and late in colorectal carcinogenesis, with the apparent exception of local recurrences.

Introduction

Both cytogenetic and molecular genetic investigations have identified acquired, nonrandom aberrations involving chromosome arm 1p, in particular its distal part, in primary colorectal tumors. Roughly 50% of cytogenetically abnormal colorectal carcinomas show deletions of 1p [1–4] and in large bowel adenomas, some investigators find 1p deletions in one-third of karyotypically abnormal cases [5]. Likewise, allelic imbalance (Al)/loss of heterozygosity (LOH) has been found at 1p loci in 20% to 80% of colorectal carcinomas and in 20% to 30% of adenomas [6–14].

The fact that loss of distal 1p has been detected as the sole cytogenetic change in both adenomas and carcinomas suggests that it is an early, perhaps primary, event in intestinal tumorigenesis [15–18]. The observation that Al at 1p loci is independent of the size of the adenoma is consistent with this view [8]. The functional importance of loss of distal 1p in colorectal tumorigenesis was demonstrated by Tanaka et al. [19], who introduced a chromosome fragment containing 1p34–36 into human colon carcinoma cells in athymic nude mice, thus achieving suppression of tumorigenicity.

The relevant gene(s) in del(1p)-mediated colorectal tumorigenesis are unknown, although both Phospholipase A2 (PLA2G2A), aflatoxin B₁-aldehyde reductase (AFAR) and tumor protein 73 (TP73) have been suggested as candidates. PLA2G2A is the human homologue of the murine Pla2g2a that has been identified as the Mom-1 gene that encodes a modifier shown to reduce tumor number and size in mice carrying a dominant Apc gene mutation [20–22]. The protein encoded by AFAR may be involved in the same signaling pathways as APC [23]. TP73 shows 63% homology with TP53 in the DNA binding domain, and p73 and p53 are believed to share functional properties, at least in part [24–26].

Several lines of evidence have suggested a pathogenetic dichotomy in colorectal carcinogenesis. Tumors in the left part of the colon and in the rectum are characteristically aneuploid with numerous chromosomal gains and losses as well as structural chromosomal rearrangements [5] affecting tumor suppressor genes and oncogenes. However, right-sided tumors are often diploid but may demonstrate genome-wide microsatellite instability (MSI) brought on by defective mismatch repair, which may lead to frameshift mutations in tumor suppressor genes and oncogenes [27]. Previous studies have suggested that 1p deletions are more common in left-sided large bowel adenomas than in right-sided ones [8,18]. Left-sided carcinomas in general show more extensive chromosomal aberrations than do right-sided tumors [5].

In contrast to the plentiful pathogenetic information on primary colorectal carcinomas, very little is known about the genetic changes that characterize the more advanced tumor lesions of this disease. We therefore undertook to compare primary, recurrent, and metastatic colorectal carcinomas, all originating from the left side of the colon or from the rectum, with regard to alterations at chromosome arm 1p.

Material and Methods

Patients and Tumor Samples

In the first part of the study, 80 fresh-frozen colorectal adenocarcinomas from 67 patients were studied, including 28 primary carcinomas, 19 local recurrences and 33 metastases (2 lung metastases, 24 liver metastases and 7 peritoneal carcinomatoses). The clinicopathologic variables are summarized in Table 1. All tumors were classified according to the World Health Organization recommendations [28]. In addition, a hematoxylin-eosin stained section was taken from 73/80 fresh-frozen samples and the percentage of tumor parenchyma cells was determined together with reevaluation of the tumor grade.

Table 1.

Clinicopathologic Data.

	Patients* (n=67)	Age Mean (Range)	Sex†	Dukes' Stage‡	Tumors§ (n=80)	Site of Primary Tumor¶	Differentiation#	% of Tumor Cells** Mean (Range)
Primary carcinomas	28	64 (24–85)	F: 14, M: 14	A: 2, B: 17, C: 6, D: 1	28	Left colon: 11, rectum: 17	L: 0, M: 23, H: 5	51% (15–100%)
Local recurrences	17	63 (32–79)	F: 7, M: 10	A: 0, B: 8, C: 7, D: 1	19	Left colon: 3, rectum: 15	L: 0, M: 15, H: 3	52% (10–80%)
Metastases	26	55 (16–74)	F: 8, M: 18	A: 0, B: 7, C: 7, D: 10	33	Left colon: 13, rectum: 19	L: 4, M: 28, H: 0	67% (10–90%)

^*Four patients are included in more than one tumor group.

^†F=female, M=male.

^‡Information about Dukes' stage was not available from two patients with primary carcinomas, one patient with local recurrence, and two patients with metastases.

^§Twenty-eight primary carcinomas from 28 patients, 19 local recurrences from 17 patients (one patient had three local recurrences), and 33 metastases from 26 patients (three patients with two liver metastases each, one patient with three liver metastases, one patient had a peritoneal carcinomatosis and a liver metastasis, and from one patient two carcinomatosis samples were analyzed).

^¶One patient, from whom a local recurrence and a metastasis were analyzed, had initially two primary carcinomas, one rectal and one in sigmoideum. From which of the two primaries the advanced tumors originated is unknown.

^#L=low, M=medium, H=high. Differentiation data was not available from one local recurrence and one metastasis.

^**Information regarding the percentage of tumor cells was not available from the frozen samples of five primary carcinomas, one local recurrence, and one

In the second part of the study, 16 formalin-fixed and paraffin-embedded primary carcinomas were analyzed and compared with the corresponding distant metastases from 16 of the patients included in the first part of the study. From two additional cases, the primary and metastatic tumors were both available as frozen specimens, giving a total of 18 primary tumor-metastasis pairs from the same number of patients.

Peripheral blood samples were available from all patients.

DNA Extraction from Frozen Tumors and Blood

DNA was extracted from frozen tumor tissue and peripheral blood leukocytes using standard phenol and chloroform extraction followed by ethanol precipitation (nucleic acid extractor, Model 340A, Applied Biosystems, Foster City, CA).

Microdissection and DNA Extraction from Archival Material

Formalin-fixed and paraffin-embedded tumor material was cut into 5-µm-thick serial sections and dried, deparaffinized, rehydrated, and given a weak hematoxylin and eosin staining. Manual microdissection of pure tumor cell populations was performed as previously described [29]. In short, an area containing 200 to 600 cells was dissected using the tip of a glass pipette and transferred to a sterile tube. Buffer [Proteinase K (0.1 mg/ml) and Tris/HCI (0.05 M, Ph 8.0)] was added and the samples were incubated over night, followed by inactivation of the proteinase K, prior to PCR-based analysis.

Microsatellite Analyses

Fourteen (CA)n microsatellite loci mapping to chromosome 1, 13 spanning the short arm, including one locus within the PLA2G2A gene, and one locus mapping to 1 q, were used to analyze the fresh-frozen primary carcinomas. Eleven of these loci were used to study the metastases and local recurrences. The following markers were analyzed (chromosome location in parentheses): D1S243 (1p36.3), D1S468 (1p36.3), D1S2845 (1p36), D1S228 (1p36), D1S2644 (1p35–36), D1S2647 (1p35–36), D1S199 (1p35–36.1), PLA2 (1p35–36.1), D1S2843 (1p34–36), D1S234 (1p34–36), D1S201 (1p32–35), D1S248 (1p13–22), D1S250 (1p11–22) and D1S305 (1cen-q21). All primer sets, except the PLA2G2A primers [30] (MedProbe, Oslo, Norway), were obtained from Research Genetics, Huntsville, AL [31]. The mapping order and the genetic distances between the loci are shown in Figure 1 [31,32].

Figure 1.

Smallest region of overlap (SRO) of 1p deletions in primary carcinomas, local recurrences, and metastases. () Homozygote; (□) retained heterozygosity; (■) Al/LOH; nd, not determined. L, liver metastasis; M, lung metastasis; P, primary carcinoma; C, carcinomatosis; R, local recurrence. The number given between the markers is the approximate distance in centi-Morgan (cM).

Most of the primer sets were run in multiplex PCR according to previously described procedures [6]. The PCR products were separated on denaturing polyacrylamide gels and visualized by autoradiography.

Archival material from primary carcinomas was investigated at three or four loci selected on the basis of the results obtained when analyzing the corresponding fresh-frozen metastases, i.e., these loci showed changes in the metastasis or helped define a breakpoint by retained heterozygosity. The microsatellite analyses were performed as for the fresh-frozen samples but with another PCR program [29].

PCR Restriction Fragment Length (PCR-RL) Analysis of TP73

An allelic polymorphism consisting of a double nucleotide substitution (G to A) and (C to T) at positions 4 and 14 of exon 2 in TP73 was used. Restriction fragment length analysis was performed after two sequential, nested PCR reactions [25]. The primer sets (Amersham Pharmacia Biotech, Lillestràm, Norway) were those described by Kaghad et al. [25]. PCR was carried out in a total volume of 50 µl containing 100 ng genomic DNA, 1xbuffer (10 mM Tris at pH 8.3, 50 mM KCl, 0.01% bovine serum albumin), 2.5 mM of each nucleotide (dATP, dCTP, dTTP and dGTP), 1 mM Mg^{2 +}, 300 nM of each of the forward and reverse primers, and 1.25 units of Taq polymerase (Perkin Elmer, Oslo, Norway). The same PCR conditions were used for both PCR reactions: 1 minute at 95°C followed by 30 cycles each of 30 seconds at 94°C, 1 minute and 30 seconds at 58°C, and 2 minutes and 50 seconds at 72°C with a 10-minute extension during the last cycle. Fifteen microliters of the resulting PCR products was Sty1 (New England BioLabs, Heidenreich, Oslo, Norway) digested at 37°C for 6 hours, run on a nondenaturing 7.5% polyacrylamide gel, and analyzed after ethidium bromide staining of the gel.

Evaluation of Allelic Imbalance (Al)/Loss of Heterozygosity (LOH)

Tumor DNA and matching blood DNA genotypes were visually and independently compared by two of the authors. Only constitutional heterozygotes were scored for Al or LOH. Al means a skewed intensity ratio between the two alleles in tumor DNA compared to blood DNA, whereas LOH is defined as the complete absence of one allele in tumor DNA compared to blood DNA. Due to the presence of normal tissue in the tumor sample and/or tumor heterogeneity, also Al most likely reflects the loss of one allele, although we cannot rule out the possibility that gain of one or more copies of chromosome 1 may also lead to Al. However, whereas gain of chromosome 1 is a rare numerical change in colorectal tumors, loss of 1p is common [33], and both the Al and LOH are described in the first part of this paper as Al and interpreted to be the result of deletions. In the second part of the study, however, in which the primary carcinomas and metastases from the same patients are compared, we do distinguish between Al and LOH. This is because only tumor parenchyma was present in the microdissected sample. Major tumor heterogeneity is also highly unlikely because only a minute area of the primary carcinoma was microdissected and analyzed.

All positive findings were confirmed by renewed analysis of a second independent PCR and electrophoresis. For the confirmation analysis of archival specimens, a new (neighboring) section (5-µm thick) was microdissected and subjected to DNA analysis.

Criteria to Determine the Smallest Region of Overlap (SRO)

Numerous Al/LOH studies of solid tumors have described SROs at several commonly deleted chromosomal regions, but rarely are the criteria used to infer the existence of an SRO made explicit. To avoid overinterpretation of data by, e.g., extrapolations to general conclusions on the basis of the genotype pattern in single tumors, we here offer a set of criteria for Al/LOH studies as well as a brief rationale for each of them. 1) The closest flanking markers on each side of the SRO must reveal retention of heterozygosity in at least two tumors. This implies that either two tumors both define each side or that 2x2 tumors define separate borders. Two is a minimum number because changes in more than one tumor are less likely to reflect random events. The number of tumors that define an SRO should also be above a certain percentage (we use 10%) of the total number of tumors exhibiting LOH. 2) If the SRO is narrowed down to a single marker, at least two contiguous markers must reveal Al in at least one tumor. This ensures that a possible mutational hot spot or a somatic mutation in a primer sequence is not misinterpreted as an SRO. Criteria 1 and 2 imply that the SRO represents a peak of Al. 3) At least two contiguous markers on each side of the SRO must show retention of heterozygosity, although not necessarily in the same tumor.

Statistical Analysis

Associations between different variables were examined using Pearson's χ² test. P values <0.05 were considered to indicate statistical significance.

When comparing individual tumor groups, patients (n=4) with tumors in more than one group were excluded from the statistical analyses. Patients (n=7) with several tumors within one group were counted once only, because all tumors from each of the seven cases showed identical 1p status. When comparing tumor site and the frequency of Al, patients (n=11) from whom more than one tumor specimen was taken were counted only once, and if the tumors from the same patient showed different 1p genotype, the patient was excluded (n=2). Finally, one case was uninformative with regard to primary tumor site. We analyzed a local recurrence and a metastasis from this patient, but it is unclear from which of two primary carcinomas (one located in the rectum and one in the sigmoideum) they originated.

Results

Frequency of Al in Primary and Advanced Colorectal Tumors

Al at one or more loci on 1p was observed in 14/28 (50%) of the primary carcinomas, in 5/19 (26%) of the local recurrences, and in 23/33 (70%) of the metastases. When patients instead of tumors within each group were compared the same frequencies of 1p loss (50%, 33%, and 64%, respectively; P=.19) were observed. Significant difference in 1p loss among the individual tumor groups was only seen between the local recurrences and the metastases (P=.04). The frequency of Al at each locus for each tumor group is illustrated in Figure 2. The most commonly altered loci were D1S2644 in the primary carcinomas (10/18) and D1S199 in the metastases (15/26). Few of the local recurrences showed loss of 1p.

Figure 2.

1p loss in primary colorectal carcinomas (crisscrossed columns), local recurrences (striped columns), and distant metastases (filled columns). ^aThe marker is analyzed only in the primary carcinomas.

Determination of SRO

The SROs identified for each of the three tumor groups (Figure 1) showed overlap with the region between markers D1S2647 and D1S2644 (orientation and distance: centromere-D1S2647-2.7 cM-D1S2644-pter) being shared by them. The SRO for the primary carcinomas was flanked by markers D1S2647 and D1S228 (defined by cases 4P, 12P, 25P, and 300P), whereas markers D1S201 and D1S228 defined the SRO in the local recurrences (cases 11R, 22R, and 78R) and D1S234 and D1S2644 defined the SRO in the metastases (cases 3L, 41L, and 48M) (Figure 3).

Figure 3.

Al within the SRO of the metastases. Al, allelic imbalance; +, retained heterozygosity; B, blood; L, liver metastasis; M, lung metastasis; SRO, smallest region of overlap.

1p Changes and Tumor Site

The clinicopathologic variables are summarized in Table 1.

No overall difference in the frequency of Al was seen between tumors originating from the rectum versus those from the left-sided colon (21/41, 51% vs 12/23, 52%). The same was the case when the two sites were compared in the individual tumor groups: primary carcinomas 8/17, 47% vs. 6/11, 55%, local recurrences 5/15, 33% vs. 0/1, 0%, and metastases 10/13, 77% vs 7/12, 58% (P=.32).

The mean percentage of tumor parenchyma cells was estimated to be 51% in the primaries, 52% in the local recurrences, and 67% in the metastases (Table 1). Among all tumors, only 6/73 (8%) had less than 30% tumor parenchyma cells.

Comparison of 1p Changes in the Primary Tumor and Metastasis from the Same Patient

The comparative study of 18 pairs of primary carcinomas and their corresponding distant metastases (Table 2, Figure 4) showed the same genotype patterns in the primary and metastasis in 12/18 cases. Nine of the 12 pairs revealed an identical pattern whereas 3/12 were interpreted as having the same, because the presence of normal DNA in the metastasis most likely masked LOH as Al and Al as retained heterozygosity. Two (cases 42 and 52) showed complete loss of one allele in the primary carcinoma and Al in the metastasis and one (case 46) showed Al in the primary carcinoma but retained heterozygosity in the metastasis. Among the remaining six cases, three (24, 37, and 39) revealed LOH at a single locus in the primary carcinoma but retained heterozygosity in the metastasis. In case 3, Al was found in the metastasis, whereas the primary carcinoma retained heterozygosity. In case 12, Al was seen in both the primary carcinoma and the metastasis, but at different loci. In case 38, finally, complete loss of one allele was seen in both the primary carcinoma and the metastasis, but different parental alleles were affected.

Table 2.

Comparison of Microdissected Primary Carcinoma (Formalin-Fixed and Paraffin-Embedded) and the Metastasis (Fresh Frozen) from the Same Patient.

Patient	Marker	Primary Carcinoma	Metastasis	Patient	Marker	Primary Carcinoma	Metastasis
3	TP73	-	-	37	D1S228	+	+
(40%)	D1S243	nd	+	(70%)	D1S201	nd	+
(50%)	D1S228	+	+		D1S248	nd	+
	D1S2644	+	+		D1S250	LOH	+
	D1S2647	+	Al
	D1S199	+	Al	38	D1S228	+	+
	PLA2	-	-	(60%)	D1S199	Al	Al
	D1S234	+	+		D1S248	LOH*	LOH*
	D1S201	-	-		D1S305	+	+
	D1S248	+	+
	D1S250	+	+	39	D2S199	-	-
	D1S305	+	Al	(80%)	D1S234	nd	Al
					D1S201	LOH	+
9	D1S228	nd	LOH
(80%)	D1S199	Al	Al	41	D1S199	LOH	LOH
	D1S234	nd	LOH	(70%)	D1S234	nd	Al
	D1S201	+	+		D1S201	-	-
12	TP73	-	-	42	D1S228	Al	Al
(20%)	D1S243	+	Al	(90%)	D1S199	LOH	Al
(70%)	D1S228	+	+		D1S201	LOH	Al
	D1S2644	Al	+		D1S248	+	+
	D1S2647	+	+
	D1S199	+	+	45	D1S199	Al	Al
	PLA2	-	-	(80%)	D1S234	nd	Al
	D1S234	-	-		D1S201	Al	Al
	D1S201	+/MSI	+/MSI		D1S248	+	+
	D1S248	-	-
	D1S250	+	+	46	D1S199	+	+
	D1S305	+	+	(80%)	D1S201	+	+
					D1S248	Al	+
13	D1S228	+	+		D1S305	+	+
(10%)	D1S199	+	+
	D1S250	+	+	48	D1S228	LOH	LOH
				(90%)	D1S199	LOH	LOH
21	D1S228	LOH	LOH		D1S250	LOH	LOH
(60%)	D1S199	LOH	LOH		D1S305	+	+
	D1S248	LOH	LOH
	D1S250	LOH	LOH	49	D1S228	+	+
				(90%)	D1S199	+	+
24	D1S228	+	+		D1S201	+	+
(50%)	D1S199	-	-		D1S305	+	+
	D1S248	LOH	+
	D1S250	+	+	52	D1S228	nd	+
				(50%)	D1S199	LOH	Al
28	D1S228	nd	Al	(80%)	D1S248	nd	+
(60%)	D1S199	Al	Al		D1S305	+	+
	D1S250	nd	Al
	D1S305	nd	+	55	D1S228	+	+
				(40%)	D1S199	+	+
					D1S305	-	-

(+) heterozygote; (-) homozygote; Al, allelic imbalance; LOH, loss of heterozygosity; MSI, microsatellite instability; nd, not determined. Percent of tumor cells in frozen biopsies are given in parentheses; in cases 3 and 12 both the primary and metastatic tumors were frozen samples.

^*At this locus different parental alleles were affected in the primary carcinoma and in the metastasis.

Figure 4.

Analysis of a primary carcinoma and a distant metastasis from the same colorectal cancer patient. (A) HE-stained section (size x40) of 24P prior to microdissection. (B) The same section of 24P after microdissection. (C) Microsatellite analyses. B, blood; P, primary carcinoma; M, lung metastasis; +, heterozygote; LOH, loss of heterozygosity.

Discussion

In our series of colorectal cancers, 1p loss was found in half of all primary carcinomas, a frequency well within the 20% to 80% reported in previous studies [7,9–14]. The broad frequency range among individual reports can be explained by chance variations due to small number of markers and/or few tumors analyzed in many studies, or biologic variation among the tumor materials may exist with regard to, e.g., the number of left-sided versus right-sided tumors. Finally, the criteria for Al/LOH have not been the same in all studies, and also the fraction of non-neoplastic cells, which may easily mask Al, may have differed from series to series.

One-third of the local recurrences showed 1p loss compared with half of the primary carcinomas and two-thirds of the metastases. The average percentage of tumor parenchyma cells in the samples did not differ among the three groups and thus cannot explain the observed differences. The comparisons are mostly group-wise because we had access to multiple lesions from the same patients in only a minority of the cases, calling for caution in the interpretation of the available data. The findings nevertheless suggest that after removal of the primary carcinoma, the remaining tumor cells with 1p loss do not enjoy any selective growth advantage locally. To our knowledge, 1p alterations in local recurrences of colorectal carcinomas have not been reported before. The high frequency of 1p loss in metastases, however, suggests that these alterations are important in the metastatic process. A limited number of colorectal liver metastases have previously been studied [7,10,34] and our results are in agreement with those of Leister et al. [7], who found LOH in five of seven liver metastases.

Our criteria for defining an SRO are strict, and consequently any region lost in only a small minority (<10%) of tumors showing AI/LOH will not be detected as an SRO. The SROs for the three tumor groups of this report were overlapping, inasmuch as the region between markers D1S2647 and D1S2644 was included in all of them. Two previous studies have defined an SRO overlapping with this region in primary colorectal carcinomas [11,35]. In the latter study, an association between the incidence of LOH in the SRO (D1S199–D1S2644) and Dukes' stages C and D was noticed, which is interesting because the SRO in the metastases of the present study completely overlaps with the region found by Matsuzaki et al. [35]. All three studies present evidence in support of the hypothesis that one or more tumor suppressor gene(s) important in colorectal carcinogenesis exist in 1 p35–36.

The candidate genes PLA2G2A (1p35–36.1) and TP73 (1p36.3) examined in the present study were seen to be proximal and distal to the SRO identified for primary carcinomas. The loss frequency of 40% for PLA2 found for both primary carcinomas and metastases is within the range (25% to 64%) previously reported for primary carcinomas [30,36,37]. Despite this relatively high frequency of loss of one PLA2 allele, the remaining allele has never been found to be mutated [30,36–38]. Although Pla2g2a, the Mom-1 locus in mice, has been shown to modify the polyp number and size in Min mice [22], this has not been seen for the human homologue, the PLA2G2A gene, in familial adenomatous polyposis (FAP) patients. Because PLA2G2A remains within the SRO of the metastases, we cannot exclude it as a potential target gene, but overall our results are better compatible with the view that complete inactivation of PLA2G2A is not a major event in colorectal carcinogenesis. The lower frequency of Al at TP73 in the primary carcinomas and metastases, the fact that TP73 is located outside the SROs, and the fact that no point mutations have been reported in TP73, except in three lung cancer cell lines [25,39–47], indicate that TP73 is relatively unimportant in colorectal tumorigenesis.

The recently cloned gene, AFAR, which is associated with hepatocellular carcinogenesis [23], is located between the markers D1S2647 and D1S199 within the SRO detected in the metastases of our series, but outside the SRO for the primaries. AFAR and PLA2G2A lie 1 cm apart, both within 1p35–36.1. It appears that of the three genes mentioned here, AFAR remains the best candidate for the functionally important target of the deletions observed distally at 1p. Finally, because several of the tumors reveal a genotype pattern consistent with a unison loss of more than one of these genes this in itself may provide the cells with a selective growth advantage.

The primary colorectal carcinoma and a distant metastasis from the same patient could be examined in 18 cases. Only one area was microdissected from each of the formalin-fixed and paraffin-embedded primary carcinomas (n=16), but because the intratumor distribution of 1 p deletions has been shown to be homogeneous in most primary colorectal carcinomas [48], we can assume that our findings are representative for the whole tumor. Nine of the 18 cases showed identical genotypes in both lesions, including three primary carcinoma-metastasis pairs in which neither lesion showed 1p loss, implying either that this genetic change is not necessary in the metastatic process, or that a micro-deletion/mutation in a 1p35–36 target gene remain undetected. Three cases showed LOH or Al in the primary and Al or retained heterozygosity in the corresponding metastasis. It is likely that the changes in each of these tumor pairs are in fact identical because the microdissected samples from the primary carcinomas contain only tumor cells, whereas in the undissected metastases, admixture of normal tissue cause LOH to be masked as Al or an Al to become undetectable. Three cases (24, 37, and 39) showed LOH in the primary carcinoma but retention of heterozygosity in the metastasis. The fraction of non-neoplastic cells was less than 50% in these metastases, which is not enough to mask a loss. Instead, this indicates that loss of 1p was not part of the metastatic process in these cases, inasmuch as the cells colonizing the liver from the primary carcinoma evidently belonged to a 1p-intact subclone.

The results from the three remaining tumor pairs support the importance of 1p change in the metastatic process. However, discordant results were seen in each pair, one (case 3) had Al in the metastasis but retention of heterozygosity in the primary carcinoma. The second case (case 12) showed Al in both the primary carcinoma and metastasis, but different loci were affected, and the last case (case 38) showed LOH in both tumors, but with involvement of different parental alleles. Although little intratumor heterogeneity for chromosome arm 1p deletions appears to characterize primary colorectal carcinomas [48], the data on metastatic lesions suggest that sometimes minor clones in the primary are the ones that colonize distant sites.

Previous knowledge of 1p loss as an early event in the adenoma-carcinoma sequence, and the fact that 1p changes are found in most carcinoma-metastases pairs, show the importance of this event also in late stages of colorectal carcinogenesis. In contrast, local recurrences do not seem to selectively develop from remaining primary tumor cells with 1p loss. Finally, the common SRO for the three groups of lesions maps a target gene(s) to the region between D1S2647 and D1S2644, distal to PLA2G2A and proximal to TP73.

Abbreviations

AFAR

aflatoxin B₁-aldehyde reductase

Al

allelic imbalance

LOH

loss of heterozygosity

MOM-1

modifier of Min-1 (mouse intestinal neoplasia)

MSI

microsatellite instability

PLA2G2A

phospholipase A2

SRO

smallest region of overlap

TP73

tumor protein 73