Sensitive Detection of K-ras Mutations Augments Diagnosis of

[CANCER RESEARCH 59, 5169–5175, October 15, 1999]

Sensitive Detection of K-rasMutations Augments Diagnosis of Colorectal CancerMetastases in the Liver

Carl C. Schimanski, Ulrich Linnemann, and Martin R. Berger 1

Unit of Toxicology and Chemotherapy, German Cancer Research Center, 69120 Heidelberg, Germany [C. C. S., M. R. B.], and Department of Surgery, Municipal Hospital ofNurnberg, 90419 Nurnberg, Germany [U. L.]

ABSTRACT

Postoperative survival of colorectal cancer patients is often delineatedby metastases spreading to the liver. Current clinical diagnostic proce-dures are unable to discover micrometastases in this organ. Our aim wasto develop a diagnostic tool for detecting micrometastases that are presentat the time of surgery. Therefore, a PCR-RFLP assay was set up trackingpoint mutations of the K-ras oncogene at codons 12 and 13, based onmismatch primers and restriction enzymesBstXI and XcmI. The detectionlimit of this assay was one mutant in one million wild-type cells. Onehundred forty-two patients with colorectal carcinoma were screened forthese mutations in tissue samples from their tumor, proximally adjacentmucosa, and liver. Of these, 67 patients (46%) were positive for a K-rasmutation, of which 58 had codon 12 and 9 had codon 13 mutations. Nopatient without a K- ras-positive tumor showed a mutation in mucosa, but11 patients with a K-ras-positive tumor (11 of 58; 19%) were found to beara K-ras mutation in their mucosa, and in 21 patients (21 of 64; 33%), aK-ras mutation was detected in liver tissue. Sequencing of all mutatedsamples revealed a 92% confirmation of PCR-RFLP results. In summary,the assay is a useful tool for detecting K-ras codon 12 and 13 mutationsand allows early proof of molecular determinants of liver metastases. Suchknowledge will improve the staging of colorectal cancer patients and couldbeneficially influence their prognosis if followed by an effective therapy.

INTRODUCTION

In Western countries, CRC2 is among the three most frequentmalignancies, being surpassed only by lung and prostate cancer inmen and by lung and breast cancer in women (1, 2). The knownhereditary syndromes of familial adenomatous polyposis and heredi-tary nonpolyposis colon cancer affect only a minority, whereas themajority of people contracting CRC (85%) appear to have no inheritedgenetic alterations and are therefore termed sporadic cases (3). Mo-lecular determinants of sporadic CRC include mutations in certaintumor suppressor genes (APC, DCC, DPC4, JV18-1,and p53) andoncogenes (K-ras) or a singlep53mutation in the case of the so-calledde novocancer (4–6). Hereditary syndromes show similar, althoughnot identical, combinations of mutated genes (4). An early eventwithin the pathogenesis of sporadic CRC is a point mutation of K-rasat codons 12, 13, or 61, which can be detected in about 40–50% ofthese tumor lesions (7–10).

Because K-rasmutations are typically located at these codons,specific and sensitive diagnostic procedures have been developed fordetecting mutant alleles in tissues and body fluids from CRC patients,such as primary tumor (11), mucosa (12), stool (13), and blood (14).From these studies, it is obvious that K-ras detection is feasible, butif applied alone, not sufficient for general CRC screening. On anindividual basis, K-ras or cytokeratin diagnostics have been used fortracing metastases of CRC in lymph nodes and bone marrow (15).However, the liver, which is most frequently affected by CRC me-

tastases, has not been included in screening for CRC micrometastases.That a greater emphasis on liver diagnostics should be applied issuggested by the fact that metastatic spread to this organ primarilydetermines the patient’s survival after extirpation of the primarytumor (16). This is supported by autopsy data indicating that themajority of patients who die of CRC show liver metastases (17).Current clinical diagnostic procedures, such as computed tomographyscans, detect metastases with a sensitivity of 94% if they exceed 1 cmin diameter (18) and only 31% if the metastases are below this size(19) but are unable to detect micrometastases.

To detect K-ras mutated liver micrometastases at the time ofsurgery, a highly sensitive PCR-RFLP assay, tracking point mutationsat codons 12 and 13 of the human K-rasgene, was modified and used.The detection limit of this assay, based on mismatch primers andrestriction enzymesBstXI and XcmI, was one mutant among onemillion wild-type cells. Tumor tissue, mucosa, and liver biopsies of142 patients were screened with this assay for point mutations atcodons 12 and 13 of the K-rasgene. Our results suggest that the assayis a useful tool for sensitive detection of K-ras codon 12 and 13mutations and allows early proof of molecular determinants of livermetastases. This method may improve clinical staging, which shouldresult in a beneficial influence on the prognosis of patients if followedby an effective therapy.

MATERIALS AND METHODS

Cell Lines. Three human CRC cell lines (HDC8, HDC63, and HDC101),established previously by Bruderleinet al. (20), were obtained at an earlypassage and cultivated under standard conditions (humidified atmosphere, 5%CO2 in air, and 37°C) in IMDM medium (15% FBS, 100 units/ml penicillin,and 100mg/ml streptomycin; Sigma, Munich, Germany). The original char-acterization of these cell lines was as follows. HDC8 cells originated from therectal cancer of a 81-year-old Caucasian male, classified as Dukes’ C(T3NxMx). HDC63 cells were taken and cultivated from the colon cancer of a60-year-old Caucasian male classified as Dukes’ D (T3NxMx). HDC101cellswere taken from the sigmoid carcinoma of a 63-year-old Caucasian malepatient classified as Dukes’ C (T2NxMx). The K-rasstatus of all three cell lineswas determined by dot blot hybridization with oligomeres specific for everyknown K-ras mutation. According to this method, HDC63 cells revealedwild-type status for K-ras, HDC8 cells were found mutated in codon 12(GGT3GAT), and HDC101 cells contained a codon 13 mutation(GGC3GAC; Ref. 21). To obtain appropriate ratios of wild-type and mutantcells, HDC8 and HDC101 cells were diluted with HDC63 cells to cell mixturesranging from 1:0 to 1:106 K-ras mutated to wild-type cells. These mixtureswere used to determine the limit of sensitivity of the PCR-RFLP assay.

Tissue Source and Storage.Tumor tissue (see Fig. 1 for scheme oflocation), mucosa, and liver biopsies were obtained from each of 142 patientsundergoing elective surgery for CRC who had given their informed consent.The tumor tissue originated from the center of the tumor, whereas the mucosawas taken from the proximal margin of the resectate, which was macroscop-ically free of disease. Trucut biopsies were taken from the left and right liverlobes, respectively. Unless for anatomical or surgical reasons, these biopsiesoriginated from normal-appearing tissue of liver segments 3 (left lobe) and 5(right lobe). Separate biopsies were used for histological and PCR-RFLPanalyses, respectively. Control liver tissue was obtained from 16 patients whounderwent a partial liver resection because of nonmalignant diseases. Alltissues were stored in cryovials, shock frozen in liquid nitrogen immediatelyafter extirpation, and stored at280°C until further processing.

Received 4/5/99; accepted 8/18/99.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby markedadvertisementin accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1 To whom requests for reprints should be addressed, at Unit of Toxicology andChemotherapy, German Cancer Research Center, INF 280, 69120 Heidelberg, Germany.Phone: 49-6221-423310; Fax: 49-6221-423313; E-mail: [email protected].

2 The abbreviations used are: CRC, colorectal cancer; TNM, Tumor-Node-Metastasis.

5169

Research. on April 10, 2019. © 1999 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

DNA Isolation. DNA isolation was performed using the Qiagen Blood &Cell Culture DNA kit according to the manufacturer’s recommendations(Qiagen, Hilden, Germany). Tissues weighing.100 mg were pulverized witha dismembrator (Mikro-Dismembrator II; B. Braun, Melsungen, Germany)before lysis, and tissues weighing,100 mg were homogenized in lysis bufferusing a 5-ml homogenisator. The amount of isolated DNA was determinedspectrophotometrically (Gene Quant II; Pharmacia, Freiburg, Germany).

Detection of K-ras Mutations. Mutations at K-ras codons 12 and 13 weredetected from all samples by a highly sensitive PCR-RFLP assay (Fig. 2); 300ng of DNA were used as template for the first PCR, which consisted of a 50-mlvolume containing Taq DNA polymerase (1.9 units; Sigma), deoxynucleotidetriphosphates (0.2mM; Sigma), reaction buffer (5ml; Sigma), and the oligo-nucleotide primers Ras A (sense; 59-ACTGAATATAAACTTGTGGTCCAT-GGAGCT-39) and Ras B (antisense; 59-TTATCTGTATCAAAGAATGGTC-

Fig. 1. Scheme of six human colorectal segments(1, cecum; 2, ascending colon;3, transversal colon;4,descending colon;5, sigma; 6, rectum) indicating thedistribution of carcinomas (n 5 146) according to theirK-ras status (F, wild type;,, mutation at codon 12;l,mutation at codon 13).

Fig. 2. Scheme of two-step PCR-RFLP assayfor detecting K-rasmutations at codons 12 and 13with restriction enzymesBstXI or XcmI, respec-tively.

5170

COLORECTAL CANCER LIVER MICROMETASTASES AND PCR-RFLP



CTGCACCA-39; 0.2 mM each). The first PCR generated an amplicon of166-bp length. For amplification, a DNA Engine PTC200 (MJ Research,Watertown, MA) thermocycler was used. Cycling conditions of the first PCRwere as follows: initial denaturation (4 min at 95°C), followed by 30 cycles ofdenaturation (1 min at 94°C), annealing (1 min at 50°C), and elongation (2 minat 72°C). After the last cycle, a final extension (5 min at 72°C) was added, andthereafter, the samples were kept at 4°C. The Ras A (sense) primer (mismatchbases are underlined) had been designed to introduce a restriction site forBstXIandXcmI into the wild-type amplicon. Because of this altered sequence,BstXI(59-CCANNNNNNTGG-39) restricted the resulting amplicon if the first twobases of codon 12 (underlined bases) were wild type. Similarly,XcmI (59-CCANNNNNNNNNTGG-39) cut the amplicon only, if the first two bases ofcodon 13 (underlined bases) were wild type.

For the first restriction, 10ml from the first PCR reaction were digested withfive units of eitherBstXI or XcmI in a total volume of 20ml, according to themanufacturer’s recommendations (New England Biolabs, Schwalbach, Ger-many). Threeml of this first digest were used as template for the second PCR,in which primer Ras C (antisense; 59-GGATGGTCCTCCACCAGTAATAT-GGATATTA-39) was used instead of Ras B (39), thus creating a restriction sitein both mutant and wild-type amplicons for enzymesBstXI andXcmI. ThePCR was run under the same conditions as the first PCR for 32 cycles. Becauseof the nested antisense primer (Ras C), the second PCR generated a fragmentof 152 bp. For the second restriction, 10ml of the second PCR were digestedwith five units of eitherBstXI or XcmI in a volume of 40ml, as described inthe product information of the manufacturer.

Twenty-twoml of the product were run on a 5% polyacrylamide gel (Roth,Karlsruhe, Germany), stained by ethidium bromide for 1 min, and analyzedunder UV light by a video densitometer (Herolab, Wiesloch, Germany).Amplicons of mutated DNA were cut only once into fragments of 134 and 18bp, whereas amplicons of wild-type DNA were cut twice into bands of 106-,28-, and 18-bp length.

DNA Sequencing.Bands of 134-bp length, indicating mutated DNA, wereexcised from 3% agarose gels. The amplicons were purified using the HighPure PCR Product Purification kit (Boehringer-Mannheim, Mannheim, Ger-many). Five ng of the purified amplicon were used for sequencing, which wasperformed with the Big Dye RR Terminator reaction (ABI, Weiterstadt,Germany), according to the recommendations of the manufacturer. New prim-ers were used for this reaction, resulting in a fragment of 110 bp (RasSeq-39,59-ACCTCTATTGTTGGATCATA-39) and 141 bp (RasSeq-59, 59-GAATATAAACTTGTGGTCCA-39). The product was run on a 5% polyacryl-amide gel in an ABI 373A Sequencer (ABI) and analyzed for point mutationsof the respective amplicon. Samples that were positive by PCR-RFLP butnegative by sequencing were retested up to two times.

Controls. DNA from cell lines HDC8, HDC63, and HDC101 was used asstandard to control the efficacy of the restriction enzymes. Autoclaved, doubledeionized water was used as negative control.

Statistics. The age of patients was compared by calculating the mean andSD of the respective subgroup. In addition, the nonparametric Wilcoxon test

was applied (22). The contingency test,x2 test, and where appropriate theCochran-Armitage trend test were used to compare all other patient and tumorcharacteristics by group.

RESULTS

Detection Limit of the PCR-RFLP Assay. The detection limit ofthe PCR-RFLP assay was determined by using serial dilutions ofK-ras mutant in wild-type cells. As shown in Fig. 3 the band indi-cating mutant DNA (134 bp) was visible even at the highest dilution(1:106) of K-ras codon 12 mutant (HDC8) in wild-type (HDC63)cells. The same was true for K-ras codon 13 mutant (HDC101) inwild-type (HDC63) cells (data not shown).

Patient Data. Tumor tissue, mucosa samples, and liver biopsies of142 patients who underwent elective surgery for CRC were screenedfor point mutations at codons 12 and 13 of the human K-ras gene(Table 1). From these 142 patients, 146 colorectal cancers wereinvestigated (including three CRCs from one female and two CRCsfrom two female patients, respectively). Patients were closely distrib-uted regarding their gender (72 males and 70 females). The 146 CRCsoriginated from six colorectal segments (Fig. 1) with varying frequen-cies, as shown by the fact that 51% of all tumors were located in thesigmoid colon and rectum. There was no significant difference intumor origin between the two sexes. Histopathological gradinggrouped most of these tumors into G2 (82%). Again, there was nosex-related difference between the histological grades observed. Ac-cording to the TNM classification, fewer tumors were classified T1,T2, or T4 than T3, which was diagnosed in 72% of all tumors.Interestingly, when the tumor size was divided by gender, there wasa tendency for females to present with larger tumors and than males.The majority of patients showed no lymph node involvement (56%)and no distant metastases (77%), according to histological and clinicalevaluation. There was no sex related difference regarding N and Mstatus.

K-ras Mutation Status versusTumor and Patient Characteris-tics. The growth of CRCs within the six colorectal segmentsshowed no correlation with K-ras mutation status, except fortumors of the descending colon, which harbored a significantlyhigher ratio of K-ras codon 13 mutations over wild-type than all

Fig. 3. Polyacrylamide gel stained with ethidium bromide.Upper bands(134 bp),mutant;lower bands(106 bp), wild-type alleles. A mutant band is detected, even at thehighest dilution of K-rascodon 12 (HDC 8) mutant in wild-type (HDC 63) cells.ddH2O,double-deionized water.

Table 1 Patient and tumor characteristics specified by sex

Females (%) Males (%) Total (%)

No. of patients 70 (49) 72 (51) 142No. of tumors 74a (51) 72 (49) 146Colorectal segment

Cecum (1)b 6 (8) 8 (11) 14 (10)Ascending colon (2)b 14 (19) 13 (18) 27 (18)Transversal colon (3)b 12 (16) 12 (17) 24 (16)Descending colon (4)b 4 (6) 3 (4) 7 (5)Sigma (5)b 23 (31) 21 (29) 44 (30)Rectum (6)b 15 (20) 15 (21) 30 (21)

pTNM classificationT1 2 (3)c 4 (6)c 6 (4)T2 7 (10)c 11 (15)c 18 (12)T3 54 (73)c 51 (71)c 105 (72)T4 11 (15)c 6 (8)c 17 (12)N0 41 (55) 41 (57) 82 (56)N1–3 33 (45) 31 (43) 64 (44)M0 54 (77) 56 (78) 110 (77)M1 16 (23) 16 (22) 32 (23)

GradingG1 2 (3) 4 (5) 6 (4)G2 64 (86) 56 (78) 120 (82)G3 8 (11) 12 (17) 20 (14)G4

a One patient with three and two patients with two colorectal cancers.b For anatomical correlation, see Fig. 1.c Comparing T status, females tended to present more often with large tumors than

males (P5 0.055).

5171




other segments (Table 2;P 5 0.012). This tendency is even moreevident, if the ratio of K-ras codon 13 mutations:K-ras codon12/13 wild-type is compared (P 5 0.007). According to the TNMclassification, K-ras codon 12 and 13 mutations prevailed overK-ras wild-type in tumors classified as T3 as compared with T2(P 5 0.026). In contrast, the appearance of local (N) or distant (M)metastases and the histological grading of tumors was not relatedto K-ras mutations. Interestingly, female CRC patients tended tobear more K-ras codon 12 mutations (59%) than male patients(41%; P 5 0.058). Another difference became obvious whenanalyzing the age of patients. CRCs were detected in male patients(66 years) 5 years earlier than in females (71 years;P 5 0.002).The difference was influenced by K-ras status; female K-rasmutant codon 12 bearers tended to be diagnosed with CRC at anearlier age (21.7 years), whereas tumors with K-ras codon 12wild-type were diagnosed at an higher age than average (11.6years).

Male patients however, showed a nonsignificant inverse trend. Forthis reason, a comparison between male and female patients withK-ras mutant codon 12 status did not reveal any significant differencein age (P 5 0.56), whereas the respective comparison between maleand female K-ras codon 12 wild-type bearers showed a highly sig-nificant difference in age (P , 0.001).

Mutations Found by PCR-RFLP. For determining the reliabilityof the PCR-RFLP assay used, the first 40 patients were reexaminedwith enzymesBstXI and XcmI. In addition, a second restrictionenzyme (MvaI) was applied to control for codon 12 mutations. In thisseries, identical results were obtained from reexaminations, and a 90%

concordance was seen whenBstXI andMvaI results were compared.3

Sixty-seven of the 146 tumors (46%) were mutated at the K-rasoncogene according to the PCR-RFLP assay (Table 3; Fig. 4). Com-paring codons 12 and 13, codon 12 mutations (87%) were foundsignificantly more frequent than codon 13 mutations (13%;P , 0.001). Among the mucosa samples, 11 mutations at K-rascodon12 were observed. K-ras mutations of the mucosa were found only ifthe respective CRC was mutated as well. Liver biopsies were positivefor a K-ras mutation in 22 cases (Table 4). One of these 22 mutationswas found in liver tissue from a patient whose CRC was negative fora K-ras mutation, whereas the other 21 mutations were associatedwith a CRC that was positive for K-ras mutations. The majority ofthese 21 mutations (20) was related to codon 12 (95%) and only one(5%) to codon 13 (P , 0.001). In addition, the ratio of codon 12versuscodon 13 mutations was higher in liver than in tumor tissue(P , 0.001). We compared the two liver lobes and found K-rasmutations in both lobes for 5 cases, in the right lobe for 10 cases, andin the left lobe for 6 cases; thus, concordant results were found in24%. Three of the five patients with bilobal K-ras mutations werediagnosed with clinically detectable liver involvement. No mutationswere found in liver samples of control patients.

Mutations Determined by Sequencing.From a total of 99 bandsof 134-bp length, indicating mutated DNA, 95 were sequenced suc-cessfully after gel extraction (Table 3). The vast majority of mutatedbands (63 of 87) was found in tumor tissue, in which codon 12

3 C. C. Schimanski, H. U. Linnemann, and M. R. Berger. Comparison of PCR-RFLPassays detecting point mutations of the human K-rasoncogene, submitted for publication.

Table 2 Tumor and patient characteristics specified by K-ras mutation status

Wild typen (%)a

No. of codon 12mutations (%)b

No. of codon 13mutations (%)c

All mutationsn (%)d

Totaln (%)e

Number of tumors 79 (54) 58 (40) 9 (6) 67 (46) 146Colorectal segment

Cecum 9 (11) 5 (9) 5 (8) 14 (10)Ascending colon 16 (20) 9 (16) 2 (22) 11 (16) 27 (18)Transversal colon 12 (15) 10 (17) 2 (22) 12 (18) 24 (16)Descending colon 2 (3) 3 (5) 2 (22)f,g 5 (8) 7 (5)Sigma 25 (32) 17 (29) 2 (22) 19 (28) 44 (30)Rectum 15 (19) 14 (24) 1 (12) 15 (22) 30 (21)

pTNM classificationT1 3 (4) 2 (3) 1 (11) 3 (5) 6 (4)T2 14 (18)h 4 (7) 4 (6)h 18 (12)T3 52 (66)h 45 (78) 8 (89) 53 (79)h 105 (72)T4 10 (13) 7 (12) 7 (10) 17 (12)N0 46 (58) 30 (52) 6 (67) 36 (54) 82 (56)N1–3 33 (42) 28 (48) 3 (33) 31 (46) 64 (44)M0

i 61/60 (77) 45/43 (77) 8/7 (88) 53/50 (78) 114/110 (77)M1

i 18/18 (23) 13/13 (23) 1/1 (12) 14/14 (22) 32/32 (23)Grading

G1 4 (5) 2 (4) 2 (3) 6 (4)G2 63 (80) 49 (84) 8 (89) 57 (85) 120 (82)G3 12 (15) 7 (12) 1 (11) 8 (12) 20 (14)G4

SexMalesi 44/44 (56)j 23/23 (41)j 5/5 (63) 28/28 (44) 72/72 (51)Femalesi 35/34 (44)j 35/33 (59)j 4/3 (37) 39/36 (56) 74/70 (49)

Age (mean6 SD) 68.4 (11.1) 68.8 (9.2) 68.2 (8.2) 68.7 (9.1) 68.5 (10.0)Males 64.9k (11.1) 68.0 (11.1) 65.4k (9.3) 67.6 (8.0) 66.0l (10.1)Females 72.7k (9.5) 69.3 (7.5) 71.8k (4.3) 69.51 (9.7) 71.0l (9.7)

a Percentage related to all K-ras wild-type colorectal cancers.b Percentage related to all K-ras codon 12 mutated colorectal cancers.c Percentage related to all K-ras codon 13 mutated colorectal cancers.d Percentage related to all K-ras mutated colorectal cancers.e Percentage related to all colorectal cancers.f Ratio of K-rascodon 13 mutantversusK-ras codon 12/13 wild-type is higher for tumors of the descending colon than elsewhere (P 5 0.007).g Ratio of K-rascodon 13 mutantversuswild-type is higher for tumors of the descending colon than elsewhere (P 5 0.012).h Ratio of K-rasmutantversuswild type is higher in T3 than in T2 tumors (P5 0.026).i Tumor number/patient number (%).j K-ras codon 12 mutations tended to be found more frequently in females than in males (P 5 0.058).k Male patients with K-ras codon 12 wild-type tumors were diagnosed earlier than female patients (P , 0.001).l Male patients were diagnosed earlier with CRC than female patients (P 5 0.002).

5172




mutations (54 of 65) dominated over codon 13 mutations (9 of 65;P , 0.001) as predicted by PCR-RFLP. Within codon 12 mutations,a G3C transversion of the second base (glycine3alanine) was mostprevalent (45%;P 5 0.014versusGAT). The second most prevalentwas the transition mutation G3A (22%) of the second base of codon12 (glycine3aspartic acid), and these two mutations representedtwo-thirds of all K-ras codon 12 mutations (P , 0.001). Regardingcodon 13, only transition mutations of the second base (G3A;glycine3aspartic acid) were found. Mucosa tissue harbored only 9%(8 of 87) of all mutations, with G3A transition mutations mostprevalent. By comparing the mutational spectra of mucosa and tumortissues, there was an inverse ranking of the two most frequent codon12 mutations, and their ratios tended to be significantly different(P 5 0.059). No codon 13 mutations were found in mucosa tissue.Liver biopsies harbored 20% (17 of 87) of all mutations, with G3Atransition mutations being most frequent (7 of 17), followed by G3Ctransversions (6 of 17). The ratio of GATversusall other codon 12mutations in liver tissue was higher than the respective ratio in tumortissue (P5 0.06). As expected, the only codon 13 mutation in liversamples was a G3A transition mutation of the second base.

Metastases Detection by Clinical Proceduresversus PCR-RFLP. The distribution of 32 metastases detected by clinical meansversustumor stage is given in Table 5. As expected, the percentage ofhematogenic liver metastases increased from 24% (T3) to 41% (T4).When the subgroup of K-ras codon 12/13 mutated tumors is consid-ered, the respective percentage of metastases increased from 23% (T3)to only 29% (T4). However, the hepatic K-ras mutations detected byPCR-RFLP decreased from 75% (T2) to 14% (T4). We compared theeffectiveness of the two methods in patients with K-ras mutatedtumors, and only 7 of 14 clinically detected cases with liver metas-tases were verified by PCR-RFLP. On the other hand, PCR-RFLPadded 14 new cases with presumably micrometastastatic spread to theliver to those detected by clinical means.

DISCUSSION

This prospective study was performed in an unselected series of142 patients with CRCs who underwent surgery with curative inten-tion. Sensitive detection of mutated K-ras by RFLP-PCR was used toimprove the identification of cases with liver involvement in the

Fig. 4. Polyacrylamide gel stained with ethidium bromide. Tumor, mucosa, and liver (LL, left lobe;RL, right lobe) tissues of six patients (nos. 68–73) were screened for K-rascodon12 mutations. The tumors of patients 72 and 73 and all liver samples of patient 72 (one sample from the right liver lobe and two separate samples from the left liver lobe) contain aK-ras codon 12 mutation.ddH2O, double-deionized water.

Table 3 Mutational spectrum in tumor, mucosa, and liver samples

Tumor (%) Mucosa (%) Liver (%)

Mutations found by PCR-RFLPAll 67/146 (46%) 11/67 (16%) 21/64a (33%)Codon 12 mutations 58/67 (87%)b–d 11/11 (100%)b 20/21 (95%)b–d

Codon 13 mutations 9/67 (13%)b–d 1/21 (5%)b–d

Mutations determined by sequencingCodon 12

GCT (alanine) 24 (45)e–h 2 (25)e,g 6 (38)e,h

GAT (aspartic acid) 12 (22)e–h 5 (63)e,g 7 (44)e,h

AGT (serine) 7 (13)e,h 1 (06)e,h

TGT (cystine) 5 (09)e,h 1 (12)e 1 (06)e,h

GTT (valine) 4 (07)e,h 1 (06)e,h

CGT (arginine) 2 (04)e,h

Sum of codon 12 mutated sequences 54 8 16Wild type (GGT; glycine) 2 3 2No amplification 2 2

Codon 13GAC 8 (100)i 1 (100)i

Sum of codon 13 mutated sequences 8 1Wild type 1No amplification

a Sixty-seven K-rasmutated tumors were found in 64 patients.b Percentage related to all mutations detected by PCR-RFLP.c Codon 12 mutations were found more frequently than codon 13 mutations in tumor as well as in liver tissues (P , 0.001, respectively).d Ratio of codon 12versuscodon 13 mutations is higher in liver biopsies than in tumor tissue (P , 0.001).e Percentage related to all codon 12 mutations for which a sequence could be obtained.f Within all mutations detected in tumors, GGT3GCT is more frequent than GGT3GAT (P 5 0.014).g Ratio of GGT3GAT versusGGT3GCT is higher in mucosa than in tumor tissues (P 5 0.059).h Within all mutations detected in liver biopsies or tumor tissue, the ratio of GGT3GAT versusall other mutations is higher in liver than in tumor tissue (P 5 0.06).i Percentage related to all codon 13 mutations for which a sequence could be obtained.

5173




subset of K-ras-positive CRC patients. The incidence of K-ras codon12 and 13 mutations in tumor tissue was 46% and is in agreement withprevious studies reporting incidences of 40–50% (7–9, 23). Remark-ably, K-ras mutants were found more often in tumors staged T3/T4

(49%) than in T1/T2 (29%), thus indicating a preference for tumorswith increased local aggressiveness. These results correspond to thoseof Finkelsteinet al. (24), who reported a 30% incidence of K-rasmutations in T1/T2 tumors and a 40% incidence in T3/T4 tumors. Inaddition, tumors of female patients were more often positive for aK-ras mutation (53%) than their male counterparts (39%). A similarlyincreased prevalence of K-ras mutations in females was reported byNelson et al. (25) for patients with non-small cell lung cancer.Whether adenocarcinomas of the colon and lung are more susceptibleto sex-related influences than other tumors remains to be investigated.In addition, if female CRC patients harbored a K-rasmutation in theirtumors, they were diagnosed 3.3 years earlier than those with codon12 wild-type tumors. For this tendency, the significant difference inage at diagnosis between male and female patients was lost within thesubgroup of patients with K-rascodon 12 mutant tumors. Because thedistribution of K-ras codon 12 mutated tumors does not reflect thesex-related difference in age found between male and female patientswith K-ras codon 12 wild-type tumors, we hypothesize that thegrowth of K-ras codon 12 mutated tumors is less dependent onsex-related factors. Mutation analysis by sequencing showed GCT andGAT to be the most prevalent genetic alterations in tumor tissue. This

is in partial contrast to other studies based on sufficient cases, whichfound GAT but not GCT to precede in number the other mutations (7,24, 26). The discrepancy could be attributed to differences in themethods used or in the life style of the patients.

Our PCR-RFLP analysis of mucosa samples revealed an 8% inci-dence of K-rasmutations. This differs from recent publications thatdescribed a prevalence of 25–54% K-ras mutants in mucosa (27–29)and is more similar to earlier studies in which incidences weredescribed ranging from 5 to 18% (30, 31). The studies describing ahigh prevalence, however, were based on relatively low case numbers(n # 20) in combination with several mucosa samples per patient andthus could overestimate the true prevalence by chance. In our mucosamaterial, mutations were found only when the respective CRCs har-bored a mutation, which was in each case identical with that in thetumor. A larger study based on several mucosa samples per patientwill provide more information as to the value of multiple mucosasamples taken per patient.

In liver samples, PCR-RFLP analysis identified K-ras mutations in21 of those 64 patients (33%) who were diagnosed with one (or more)K-ras-positive CRCs (n 5 67). As for mucosa, the sequences of thesemutations corresponded to their primary tumor. In one case, a K-rascodon 12 mutation was found in liver tissue, together with a K-raswild-type CRC. This is indicative either of a mutation that occurredafter the metastatic spread of CRC, of an additional neoplastic processof unknown origin, or of a false-positive result.

Bilobal involvement probably is a marker of advanced disease, ascan be estimated from the total number of patients (n 5 5) that splitsrather with (3 of 7; 43%) than without (2 of 14; 14%) clinicallydetectable liver involvement (Table 4). For a more stringent evalua-tion of this relationship, however, or of the bearing of codon 12 and13 mutations in liver tissue, the number of patients is not largeenough. Regarding the mutation spectrum found in liver tissue, GATand GCT mutations of K-ras codon 12 were most prevalent, asalready observed in tumor and mucosa tissues. However, CRCs con-taining the codon 12 mutation GAT showed a distinctively highermetastatic efficiency (7 of 12; 58%) as compared with CRCs harbor-ing the codon 12 mutation GCT (6 of 24; 25%). Our observation, thatthe GAT mutation of K-ras codon 12 is most prevalent in livermicrometastases, is corroborated by the observation of Finkelsteinetal. (20), who found seven GAT mutations within 13 solid livermetastases (54%).

On the basis of the assumption that liver DNA containing the samemutation as the respective CRC is indicative of metastatic CRC cellsin the liver, we compared the efficiency of clinical and PCR-RFLPdiagnostic procedures. Clinical diagnostic means identified 32 caseswith metastases of the liver among 142 patients (23%). Fourteen ofthese (44%) were derived from a CRC, which contained a K-rasmutation. PCR-RFLP indicated metastatic CRC cells in the liver for21 cases within the 64 patients harboring a K-ras mutant tumor (21 of64; 33%). When comparing the two subgroups of patients with liverinvolvement identified by clinical and PCR-RFLP means, it is evidentthat clinically detected metastases originated from T3/T4 tumor stages

Table 4 List of all patients with PCR-RFLP K-ras-positive liver tissue

Caseno.

Patientno. Sexa Age Tb Nb Mb

Origin oftumorc

Sequence of K-rasin

LobeinvolvementfTumor Liver

1 142 2 72 2 0 0 5 CGT12 CGT12 BL2 47 1 63 2 0 0 6 GAT12 GAT12 OL3 112 1 77 2 2 0 6 WTd WT OL4 65 2 69 3 0 0 1 GAT12 GAT12 OL5 20 2 61 3 2 1 2 GAT12 GAT12 BL6 67 1 76 3 1 0 3 GAT12 GAT12 OL7 77 1 75 3 3 1 3 GAC13 GAC13 OL8 121 2 69 3 1 1 3 GCT12 GCT12 OL9 127 2 66 3 1 0 3 GCT12 GCT12 OL

10 141 1 76 3 1 0 3 WT WT OL11 5 1 78 3 0 0 3 TGT12 TGT12 OL12 41 1 66 3 0 0 4 AFe AF OL13 72 2 58 3 1 1 5 GCT12 GCT12 BL14 28 1 66 3 1 1 5 GAT12 GAT12 OL15 32 1 89 3 3 1 5 GAT12 GAT12 BL16 126 2 63 3 0 0 5 GCT12 GCT12 OL17 139 1 82 3 0 0 5 GCT12 GCT12 BL18 23 1 71 3 2 1 6 GTT12 GTT12 OL19 18 1 43 3 0 0 6 GCT12 GCT12 OL20 129 1 67 3 2 0 6 GAT12 GAT12 OL21 38 2 62 4 0 0 4 AF AF OL1 66 1 78 3 0 0 4 GCT12 OLa Number 1 codes for female; number 2 codes for male patients.b Tumor classification according to the TNM system.c 1, cecum; 2, ascending colon; 3, transversal colon; 4, descending colon; 5, sigma; 6,

rectum.d Wild-type.e Amplification failed.f OL, one lobe; BL, both lobes.

Table 5 Comparison of metastases frequency detected by clinical procedures versus PCR-RFLP

Tumorstage

All tumorsn (%)

Cases with metastasesin the liver n (%)

K-ras codon 12/13 mutated colorectal tumors

n (%)Cases with metastases

in the liver n (%)aCases with K-rasmutations

in the liver n (%)b

T1 6 (100) 3 (100)T2 18 (100) 4 (100) 3 (75)T3 105 (100) 25 (24) 53 (100) 12 (23) 17 (32)T4 17 (100) 7 (41) 7 (100) 2 (29) 1 (14)

a Metastases detected clinically at time of surgery.b Micrometastases detected by PCR-RFLP assay.

5174




only, whereas PCR-RFLP detected liver involvement of patients withT2-staged tumors (3 of 21; 14%), as well. It has to be noted, however,that PCR-RFLP did not recognize all clinically identified metastasesof K-ras-positive tumors; only 7 of those 14 (50%) were confirmed byPCR-RFLP. We hypothesize that this difference could be attributableto metastatic spread prior to K-ras mutation of the primary tumor, toa K-ras-negative subclone, or to liver biopsies taken from areaswithout metastatic cells. On the other hand, the clinical diagnosticprocedures detected solid metastases only in 7 of 21 cases that werePCR-RFLP positive for K-rasmutations in liver tissue. This resultshows that the two methods supplement each other in diagnosticvalue, rather than compete for the same information. Therefore, ifK-ras-positive metastases of the liver can be discriminated moresensitively by this than with conventional methods, it would allowinitiation of an early and specific therapy, provided an adequatetherapeutic option is available.

The value of the PCR-RFLP as a diagnostic tool can be describedby comparing the number of samples found mutated by this method(n 5 99) with those confirmed by sequencing (n 5 87). Becausesequencing failed in four samples, this gives a ratio of eight falsepositives (wild type by sequencing) in 95 sequences obtained (8%).This relatively high percentage of correct predictions (92%) showsthat the PCR-RFLP assay is a valid and reliable tool for detectingK-ras mutated cells. A follow-up study should show whether theK-ras mutated cells found in liver tissue predict the outgrowth of solidliver metastases.

REFERENCES

1. Parkin, D. M., Pisani, P., and Ferlay, J. Estimates of the worldwide incidence ofeighteen major cancers in 1985. Int. J Cancer,54: 594–606, 1993.

2. Parker, S. L., Tong, T., Bolden, S., and Wingo, P. A. Cancer statistics, 1996 [seecomments]. CA Cancer J. Clin.,46: 5–27, 1996.

3. Houlston, R. S., Collins, A., Slack, J., and Morton, N. E. Dominant genes forcolorectal cancer are not rare. Ann. Hum. Genet.,56: 99–103, 1992.

4. Kinzler, K. W., and Vogelstein, B. Lessons from hereditary colorectal cancer. Cell,87: 159–170, 1996.

5. Cho, K. R., and Vogelstein, B. Genetic alterations in the adenoma-carcinoma se-quence. Cancer (Phila.),70: 1727–1731, 1992.

6. Vogelstein, B., and Kinzler, K. W. The multistep nature of cancer. Trends Genet.,9:138–141, 1993.

7. Andersen, S. N., Lovig, T., Breivik, J., Lund, E., Gaudernack, G., Meling, G. I., andRognum, T. O. K-ras mutations and prognosis in large-bowel carcinomas. Scand. J.Gastroenterol.,32: 62–69, 1997.

8. Elnatan, J., Goh, H. S., and Smith, D. R. c-Ki-RAS activation and the biologicalbehaviour of proximal and distal colonic adenocarcinomas. Eur. J. Cancer,32A:491–497, 1996.

9. Bos, J. L.ras oncogenes in human cancer: a review [published erratum appears inCancer Res.,50: 1352, 1990]. Cancer Res.,49: 4682–4689, 1989.

10. Konishi, M., Kikuchi-Yanoshita, R., Tanaka, K., Muraoka, M., Onda, A., Okumura,Y., Kishi, N., Iwama, T., Mori, T., Koike, M., Ushio, K., Chiba, M., Nomizu, S.,Konishi, F., Utsunomiya, J., and Miyaki, M. Molecular nature of colon tumors inhereditary nonpolyposis colon cancer, familial polyposis, and sporadic colon cancer[see comments]. Gastroenterology,111: 307–317, 1996.

11. Kahn, S. M., Jiang, W., Culbertson, T. A., Weinstein, I. B., Williams, G. M., Tomita,N., and Ronai, Z. Rapid and sensitive nonradioactive detection of mutant K-rasgenesvia “enriched” PCR amplification. Oncogene,6: 1079–1083, 1991.

12. Minamoto, T., Yamashita, N., Ochiai, A., Mai, M., Sugimura, T., Ronai, Z., andEsumi, H. Mutant K-rasin apparently normal mucosa of colorectal cancer patients.Its potential as a biomarker of colorectal tumorigenesis. Cancer (Phila.),75: 1520–1526, 1995.

13. Deuter, R., Pietsch, S., Hertel, S., and Muller, O. A method for preparation of fecalDNA suitable for PCR. Nucleic Acids Res.,23: 3800–3801, 1995.

14. Nakamori, S., Kameyama, M., Furukawa, H., Takeda, O., Sugai, S., Imaoka, S., andNakamura, Y. Genetic detection of colorectal cancer cells in circulation and lymphnodes. Dis. Colon Rectum,40 (Suppl.):S29–S36, 1997.

15. Weitz, J., Kienle, P., Lacroix, J., Willeke, F., Benner, A., Lehnert, T., Herfarth, C.,and von Knebel Doeberitz, M. Dissemination of tumor cells in patients undergoingsurgery for colorectal cancer. Clin. Cancer Res.,4: 343–348, 1998.

16. Petrelli, N., Gupta, B., Piedmonte, M., and Herrera, L. Morbidity and survival of liverresection for colorectal adenocarcinoma [see comments]. Dis. Colon Rectum,34:899–904, 1991.

17. Gilbert, H. A., and Kagan, A. R. Metastases: incidence, detection, and evaluationwithout histologic confirmation.In: L. Weiss (ed.), Fundamental Aspects of Metas-tasis, pp. 385–405. Amsterdam: North-Holland, 1976.

18. Carter, R., Hemingway, D., Cooke, T. G., Pickard, R., Poon, F. W., McKillop, J. A.,and McArdle, C. S. A prospective study of six methods for detection of hepaticcolorectal metastases. Ann. R. Coll. Surg. Engl.,78: 27–30, 1996.

19. Vassiliades, V. G., Foley, W. D., Alarcon, J., Lawson, T., Erickson, S., Kneeland,J. B., Steinberg, H. V., and Bernardino, M. E. Hepatic metastases: CTversusMRimaging at 1.5T. Gastrointest. Radiol.,16: 159–163, 1991.

20. Bruderlein, S., van der Bosch, K., Schlag, P., and Schwab, M. Cytogenetics and DNAamplification in colorectal cancers. Genes Chromosomes Cancer,2: 63–70, 1990.

21. Schimanski, C. C., Sutter, C., Linnemann, U., and Berger, M. R. Sampling techniqueinfluences the detection of K-ras mutations in normal appearing mucosa of colorectalcancer patients. Int. J. Oncol.,15: 391–398, 1999.

22. Dunn, O. J. Multiple comparison using rank sums. Technometrics,6: 241–252, 1964.23. Vogelstein, B., Fearon, E. R., Hamilton, S. R., Kern, S. E., Preisinger, A. C., Leppert,

M., Nakamura, Y., White, R., Smits, A. M., and Bos, J. L. Genetic alterations duringcolorectal-tumor development. N. Engl. J. Med.,319: 525–532, 1988.

24. Finkelstein, S. D., Sayegh, R., Christensen, S., and Swalsky, P. A. Genotypicclassification of colorectal adenocarcinoma. Biologic behavior correlates with K-ras-2 mutation type. Cancer (Phila.),71: 3827–3838, 1993.

25. Nelson, H. H., Christiani, D. C., Wiencke, J. K., Wain, J., Mark, E., and Kelsey, K. T.K-ras mutations and female gender in non-small cell lung cancer. Proc. Am. Assoc.Cancer Res.,40: 202, 1999.

26. Benhattar, J., Losi, L., Chaubert, P., Givel, J. C., and Costa, J. Prognostic significanceof K-ras mutations in colorectal carcinoma. Gastroenterology,104: 1044–1048,1993.

27. Minamoto, T., Esumi, H., Ochiai, A., Belitsky, G., Mai, M., Sugimura, T., and Ronai,Z. Combined analysis of microsatellite instability and K-ras mutation increasesdetection incidence of normal samples from colorectal cancer patients. Clin. CancerRes.,3: 1413–1417, 1997.

28. Minamoto, T., Yamashita, N., Ochiai, A., Mai, M., Sugimura, T., Ronai, Z., andEsumi, H. Analysis of mutant K-rasin multiple sites of normal appearing mucosa ofcolorectal cancer patients. Int. J. Oncol.,9: 911–915, 1996.

29. Zhu, D., Keohavong, P., Finkelstein, S. D., Swalsky, P., Bakker, A., Weissfeld, J.,Srivastava, S., and Whiteside, T. L. K-rasgene mutations in normal colorectal tissuesfrom K-rasmutation-positive colorectal cancer patients. Cancer Res.,57: 2485–2492,1997.

30. Ronai, Z., Luo, F. Q., Gradia, S., Hart, W. J., and Butler, R. Detection of K-rasmutation in normal and malignant colonic tissues by enriched PCR method. Int. J.Oncol.,4: 391–396, 1994.

31. Minamoto, T., Ronai, Z., Yamashita, N., Ochiai, A., Sugimura, T., Mai, M., andEsumi, H. Detection of Ki-ras mutation in non-neoplastic mucosa of Japanese patientswith colorectal cancer. Int. J. Oncol.,4: 397–401, 1994.

5175




1999;59:5169-5175. Cancer Res Carl C. Schimanski, Ulrich Linnemann and Martin R. Berger Colorectal Cancer Metastases in the Liver

Mutations Augments Diagnosis ofrasSensitive Detection of K-

Updated version

http://cancerres.aacrjournals.org/content/59/20/5169

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/59/20/5169.full#ref-list-1

This article cites 30 articles, 4 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/59/20/5169.full#related-urls

This article has been cited by 7 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/59/20/5169To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/content/59/20/5169.full#ref-list-1

http://cancerres.aacrjournals.org/content/59/20/5169.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

Sensitive Detection of K-ras Mutations Augments Diagnosis of