CORRELATIONS BETWEEN p53-PROTEIN ACCUMULATION, SERUM ANTIBODIESAND GENE MUTATION IN COLORECTAL CANCERPascal HAMMEL 1*, Karen LEROY-VIARD2, Marie-Therese CHAUMETTE2, Jacqueline VILLAUDY 3, Marie-Claude FALZONE2,Dany ROUILLARD3, Richard HAMELIN 4, Brigitte BOISSIER5 and Yorghos REMVIKOS3

1Federation Medico-Chirurgicale d’Hepato-Gastroente´rologie, Hopital Beaujon, Clichy, France2Service d’Anatomie et de Cytologie Pathologique, Hoˆpital Henri Mondor, Creteil, France3UMR 147 CNRS, Institut Curie, Paris, France4C.E.P.H., Paris, France5Service de Biochimie, Hoˆpital Henri Mondor, Creteil, France

Only half of colorectal-cancer patients elicit serum antibod-ies in response to intratumoral p53-gene mutations. Ourstudy was designed to compare cellular events (p53-proteinaccumulation and gene mutations) with the presence ofcirculating anti-p53 antibodies (p53-Ab). Thirty-five colorectal-cancer patients were studied for their intratumoral p53-protein accumulation and circulating p53-Ab. Tumour DNAwas analyzed for genomic mutations in a sub-set of 28patients. In all, 18 tumours (51.4%) were positive by immuno-histochemistry, and 17 tumour extracts were shown tocontain ‘‘mutant’’ conformation p53 protein, 16 of thembeing were concordant by both methods. Of the 28 tumourstested by DGGE, 16 contained alterations in p53 exons 5 to 8(57.1%). Of 12 tumours without detectable mutations, 10were ‘‘mutant’’-conformation-negative by immunohistochem-istry and ELISA. Paradiploid tumors presented more fre-quently wild-type p53 genes and were significantly less fre-quently immunohistochemistry- or p53-Ab-positive thanpolyploid tumors. Circulating p53-Ab were detected in theserum of 11 patients (31%). In 9/11 cases, a gene mutation wasfound in the corresponding tumour. Three of four mutationsin exon 8 and 3/3 mutations in exons 5-6 were associated withp53-Ab, in contrast with only 3/9 mutations in exon 7. Wefound good agreement in the detection of p53-gene alter-ations by different methods. However, our data suggest thatall gene mutations may not be equivalent in term of immuno-genicity. Int. J. Cancer 81:712–718, 1999.r 1999 Wiley-Liss, Inc.

Detection of p53-tumour-suppressor-gene alterations has re-sulted in intensive investigation, due to the key role of thesealterations in human carcinogenesis.p53 alterations are usuallycaused by mis-sense point mutations of the gene, but inactivationof the p53 protein may also occur through complex formations withcellular protein (Soussiet al.,1994). Schematically,p53alterationscan be assessed in 3 ways (Soussiet al., 1994): (i ) by DNAsequencing, usually preceded by semi-direct analysis using de-naturing gradient gel electrophoresis (DGGE) or single-strandconformation polymorphism; (ii ) by demonstrating intratumoralaccumulation of the p53 protein using immunohistochemistry, flowcytometry or enzyme-linked immunosorbent assay (ELISA) (Rem-vikoset al.,1992; Hall and Lane, 1994; Tominagaet al.,1997) and(iii ) by detection of circulating anti-p53 antibodies (p53-Ab)(Lubin et al., 1995a). DNA sequencing is the gold-standardmethod, but it is not feasible on a routine clinical basis. Immunohis-tochemistry is the most widely used method for assessingp53alterations in human cancers (Soussiet al., 1994). It can beperformed during routine analysis, but some mutations may notlead to p53-protein accumulation, or p53 over-expression may bedetected in the absence of mutations of thep53 gene (Hall andLane, 1994; Tominagaet al., 1997). The correlation between themolecular detection ofp53gene mutations and p53-protein accumu-lation shown by immunohistochemistry varies widely, and usuallydoes not exceed 70% (Soussiet al.,1994). We have shown that thecombination of flow cytometry (nuclear fraction) and ELISA(soluble fraction) provides an accurate characterization of thep53-protein accumulation pattern, resulting in good agreementwith the molecular analysis of gene mutations (Tominagaet al.,

1997). Compared with the methods cited above, serologicaldetection of p53-Ab is easier to perform, does not require tumourmaterial and is of potential interest in clinical practice, since it canhelp physicians in the early detection of cancers (Lubinet al.,1995b), patient monitoring (Houbierset al.,1995; Angelopoulouetal., 1997; Hammelet al., 1997) and may perhaps be a prognosticindicator (Houbierset al., 1995). The near-100% specificity ofp53-Ab for the diagnosis of cancer is well accepted (Lubinet al.,1995a,b; Houbiers et al., 1995; Angelopoulouet al., 1997;Hammelet al., 1997). In contrast, the intrinsic limit of p53-Abtesting is its relatively low sensitivity for the detection ofp53alterations (Lubinet al.,1995a). Whereas intratumoral p53-proteinaccumulation appears to be a prerequisite for the production ofp53-Ab, the exact mechanisms of the self-immunization processhave still not been clarified, and whether the type of gene mutationmay influence the production of p53-Ab is still a matter of debate.

Colorectal cancer, one of the most common malignanciesworldwide, containsp53 alterations in about 60 to 70% of cases(Soussiet al.,1994). Our group and others have shown that p53-Abare present in about 30% of colorectal-cancer patients (Houbiersetal., 1995; Angelopoulouet al.,1997; Hammelet al.,1997).

The aims of the present study were to assess (1) the correlationsbetween intratumoral p53-protein accumulation detected by immu-nochemistry and by ELISA/FCM, serum-positivity for p53-Ab andthe presence of gene mutations, and (2) whether specific mutationsare likely to favour p53-Ab production in colorectal cancer.

PATIENTS AND METHODS

We prospectively studied 35 patients (16 men and 19 women).The mean age was 65 (range 33–90). All patients were treated for ahistologically confirmed colorectal adenocarcinoma. Specimens ofcolorectal carcinoma were obtained immediately after surgery (orbefore irradiation, for some rectal cancers) and stored in liquidnitrogen at280°C. The procedures for tumour and serum samplingare shown in Figure 1. According to the Astler-Coller classification,3 patients were stage A, 2 were B1, 14 were B2, 1 was C1, 12 wereC2 and 3 were D.

Immunohistochemistryp53 immunostaining was performed as described (Hammelet

al., 1997) on 3-µm de-paraffinized sections using the monoclonalantibody (MAb) anti-p53 DO7 (Dakopatts, Glostrup, Denmark),which recognizes an epitope in the N-terminal part of the humanp53 protein between amino acids 35 and 45 (Stephenet al.,1995).The pathologist who performed the immunohistochemical analysiswas not informed of the patient’s p53-Ab status.

*Correspondence to: Service de Gastroente´rologie, Hopital Beaujon, 100boulevard du Ge´neral Leclerc, 92110 Clichy, France. Fax: (33) 1-42-70-37-84. E-mail: philippe.ruszniewski@bjn.ap-hop-paris.fr

Received 3 November 1998; Revised 9 December 1998

Int. J. Cancer:81,712–718 (1999)

r 1999 Wiley-Liss, Inc.

Publication of the International Union Against CancerPublication de l’Union Internationale Contre le Cancer

Flow cytometryDNA indices were measured as described (Remvikoset al.,1992).

Tumours were sub-divided into paradiploid and polyploid, with a cutoffvalue of the DNAindex set at 1.25. Paradiploidy has been defined on thebasis of cytogenetic data (Remvikoset al.,1997). Moreover, the use ofFCM was justified to ensure that the extracts analyzed by ELISAcontained a sufficient proportion of tumour cells.

ELISA for measurement of intratumoral p53-protein expressionThe ELISA technique has been described elsewhere (Remvikos

et al.,1992; Tominagaet al.,1997). We used anti-p53 mouse MAbsPAb 1801 (2 µg/ml) and PAb 240 (1/10) applied to 80 µl of tumourextract (soluble fraction), followed by anti-p53 rabbit CM-1polyclonal antibody (1/1000) and horseradish-peroxidase-conju-gated goat anti-rabbit IgG (Southern Biotechnology, Birmingham,AL). Thresholds for positivity were defined as described byTominagaet al.(1997).

ELISA for detection of circulating p53-AbSera of the 35 colorectal-cancer patients were collected after

diagnosis but prior to treatment, during routine blood sampling.The ELISA test used for the detection of p53-Ab in serum has beendescribed (Lubinet al., 1995a). All results were expressed as theratio between the value of the wells containing the p53 antigen andthe corresponding wells without. All analyses were performed induplicate. Some of the results of p53-Ab testing in the patients inthis series have been reported (Hammelet al.,1997).

Identification ofp53-gene mutationsDNA was prepared from frozen tissue sections by proteinase-K

digestion and phenol-chloroform extraction after hematein-eosin-

safran coloration of a control 5-µM section. Mutations in exons 5 to8 of thep53gene were detected by analysis of the electrophoreticmigration profile of amplified DNA on DGGE, as described(Hamelin et al., 1994). Mutated DNA was amplified withp53-specific oligonucleotides containing M13 primer sequences at their58 end, and sequenced on both strands with ready-reaction dye-primer sequencing kits (Applied Biosystems, Courtaboeuf, France).Tumour fragments were from primary lesions, all but one (frompatient PT34), which originated from a liver metastasis, due to lackof remaining informative material for primary cancer.

Statistical analysisThe Chi squared and Fischer’s exact tests were used for

statistical comparisons.

RESULTS

Frozen fragments were available from 35 patients who wereparticipating in a study of p53 humoral response at the time ofdiagnosis and during follow-up. DNA ploidy was performed in allcases on the nuclear fractions after homogenization in a Douncepotter (Fig. 1). Examples are shown in Figure 2a–f. One case(PT23) repeatedly presented a DNA-diploid profile (Fig. 2b)resembling that of normal mucosa, with an S-phase fraction of4.7%. Since sampling was probably inadequate, this case wasexcluded from FCM and ELISA analysis. Seventeen tumours wereparadiploid and 17 were polyploid.

Eighteen out of 35 tumours (51.4%) exhibited significantstaining by immunohistochemistry (more than 10% stained nuclei)on paraffin sections. In 16 cases, more than 50% of the tumour cells

FIGURE 1 – Methodology used for tumour and serological sampling before studying p53 alterations in a series of 35 colorectal cancer patients.

713p53 PROTEIN IN COLORECTAL CANCER

were strongly stained. In the 2 remaining cases (PT13 and PT23),mosaic patterns with scattered positive glands (range of stainedcells from 10% to 75%) were seen and considered as positive (Fig.3a,b). The presence of p53 protein in the mutant conformation wasconfirmed by ELISA using PAb 240 in 16/18 (89%) tumourextracts. One case (PT23) was not tested for the reason mentionedabove, and another case (PT28) was discordant (see Table I).

p53-gene mutations were investigated in 28 cases. As shown inTable I, 16 out of 28 tumours tested by DGGE containedp53-genealterations (57.1%), located as follows: 1 in exon 5, 2 in exon 6, 9in exon 7, and the remaining 4 cases in exon 8. In 2 cases, tumour

DNA exhibited 2 mutations (patients PT16 and PT26). Of the 16tumours with abnormal DGGE, 13 were further sequenced (TableI): 7 out of 15 sequenced mutations were located in codons 248 and273.

Seven tumours were not investigated forp53-gene mutations; 4of them (PT11, PT20, PT21 and PT33) exhibited a paradiploidDNA index, an absence of histochemical p53-protein accumula-tion, and an ELISA profile (PAb 18011, PAb 2402) stronglysuggestive of wild-type-p53-protein expression (Tominagaet al.,1997). In 3 cases (PT5, PT13 and PT32), no remaining frozentumour fragments were available for DGGE analysis.

FIGURE 2 – Examples of demonstrative DNA histograms. Normal mucosa(a), tumours of patients PT23(b), PT8 (c), PT14(d), PT4 (e) andPT29(f).

714 HAMMEL ET AL.

Out of 35 patients, 11 (31.4%) were serum-positive on the basisof their p53 ratio (from 2.1 to 65) (see Table I).

Overall, the agreement between the methods used for thedetection ofp53-gene alterations was excellent (Table I). All 14tumours exhibiting p53-protein accumulation, as detected byELISA and by immunochemistry, exhibited ap53-gene mutationwhen tested by DGGE analysis. On the other hand, 10 out of 12

tumours without detectable gene mutations were also negative forELISA-mutant conformation and for immunohistochemistry.

The correlation between p53 serology and the detection ofp53alterations using the other methods was also quite good. A genemutation was identified in all serum-positive patients but 2 (seeTable I). An analysis of the type of mutations in serum-positivepatients showed that 3/4 mutations in exon 8 and 3/3 mutations in

FIGURE 3 – (a) Immunohistochemical analysis of a tumour specimen (patient PT23) showing stronger staining (50% of positive cells)(magnification:3400). (b) Analyzis of another area of the same tumour specimen showing weaker p53 staining (about 10% of positive cells)(magnification:3250).

TABLE I – CORRELATIONS BETWEEN IMMUNOHISTOCHEMISTRY (IHC), ANTI-p53 ANTIBODIES (p53-Ab) AND GENE MUTATIONS IN THE35 COLORECTAL CANCERS STUDIED

Patient1 Site Gender Stage DNA index IHCELISA

p53-Ab RatioGene mutation

1801 240 DGGE Sequencing

PT1 (CC1) LC F D 1.65 1 1 1 1 26.4 exon 7 codon 248 (CGG= TGG)PT2 RC M B2 1 2 1 2 2 noPT3 (CC3) LC F C2 1.49 1 1 1 1 4.0 exon 6 deletion 192 (13336= 13338)PT4 LC F C2 1.12 1 1 1 2 exon 7 deletion 255 (14091= 14093)PT5 R M C2 1.74 1 1 1 2 n.d.PT6 (CC6) RC F B2 1.55 2 2 2 1 3.2 noPT7 LC M B2 2.06 1 1 1 2 exon 7 codon 248 (CGG= CAG)PT8 R F C2 1.02 1 1 1 2 exon 7 n.d.PT9 LC M A 1 2 1 2 2 noPT10 LC M B2 1.02 2 1 2 2 noPT11 RC M A 1 2 1 2 2 n.d.PT12 LC F C2 1.27 2 1 2 2 noPT13 R F C2 1.85 1 1 1 2 n.d.PT14 (CC21) RC M D 1.74 1 1 1 1 36.9 exon 8 codon 273 (CGT= TGT)PT15 LC M B2 1.30 1 1 1 2 exon 8 codon 273 (CGT= TGT)PT16 (CC24) LC F B2 1.21 1 1 1 1 58.1 exon 5 codons 177 and 181PT17 R F B2 1.19 2 1 2 2 noPT18 LC M B1 1 2 1 2 2 noPT19 (CC30) LC M B2 1.84 1 1 1 1 36.4 exon 7 codon 245 (GGC= AGC)PT20 R M C2 1 2 1 2 2 n.d.PT21 RC M A 0.94 2 1 2 2 n.d.PT22 LC M B2 1.24 2 2 1 2 noPT23 (CC43) RC F B2 n.i. 1 n.d. n.d. 1 3.2 noPT24 (CC37) R F C2 1.87 1 1 1 1 65 exon 7 codon 250 (CCC= CTC)PT25 (CC38) LC M B1 1.58 1 1 1 1 11.1 exon 6 codon 220 (TAT= TGT)PT26 LC F C2 1.56 1 1 1 2 exon 7 codons 241 and 248PT27 LC F B2 1.17 2 1 2 2 noPT28 R F C2 1.52 1 1 2 2 exon 7 n.d.PT29 RC F C2 1.08 1 1 1 2 exon 7 codon 248 (CGG= TGG)PT30 (CC47) R M C2 1.45 2 1 2 1 3.4 exon 8 n.d.PT31 R M B2 0.96 2 2 2 2 noPT32 R F B2 1.78 2 1 2 2 n.d.PT33 R F C1 1 2 1 2 2 n.d.PT34 RC F D 1.38 2 2 2 2 noPT35 (CC53) LC F B2 1.1 1 1 1 1 2.7 exon 8 codon 273 (CGT= TGT)

LC, left colon; RC, right colon; R, rectum; M, male; F, female; n.i., not informative; n.d., not done.–1Numbers in parentheses are patient codesused in our earlier report (Hammelet al.,1977).

715p53 PROTEIN IN COLORECTAL CANCER

FIGURE 4 – Electrophoretic migration profile of amplified DNA from tumours of patients PT3, PT10, PT16 and PT19.(a) PCR exploring thedistal region of p53 exon 5. M, mutated DNA; N, normal DNA used as controls. PT 16 DNA exhibits numerous abnormal bands (mis-sensemutations of codons 177 and 181).(b) PCR exploring p53 exon 6. PT3 DNA exhibits an abnormal profile (3nt deletion preserving p53 openreading frame).(c) PCR exploring p53 exon 7. PT19 DNA exhibits an abnormal profile. PT19 exon-7 sequence is shown on the left (mis-sensemutation of codon 245.

716 HAMMEL ET AL.

exons 5-6 were associated with p53-Ab, in contrast with only 3/9mutations in exon 7. If exon 7 mutations are compared with therest, the difference tends to be significant (p 5 0.06) despite thesmall number of cases tested. Moreover, 2/3 of the mutations incodon 273 were immunogenic, compared with only 1/4 of those incodon 248. One patient (PT30) had both p53-Ab and a genemutation located in exon 8, while immunohistochemistry wasnegative. Unfortunately, we could not sequence the gene because ofthe low quality of the available tumour samples. Examples ofDDGE results are shown in Figure 4.

Patient PT6 had circulating p53-Ab as a single abnormality inthe p53 system. The high p53-Ab ratio was confirmed in severalsamples, and remained unchanged during patient follow up (Ham-mel et al., 1997). Another patient (PT23) had weakly positivep53-Ab levels (p53 ratio/control: 2.2 to 2.8) and scattered positiveglands upon immunohistochemical analysis, whereas no mutationswere found on this tumour specimen. Fourteen months after tumourresection, the patient was in good condition and p53-Ab becameundetectable (Hammelet al.,1997).

Of 22 tumours located in the distal colon, 14 had mutations, incontrast to only 2 of 6 tumours located in the proximal colon, butthe difference was not significant. The cancer site did not influencethe prevalence of p53-Ab (8/27vs.3/8, ns).

DNA-ploidy correlated top53 alterations whatever the methodused. Thus, DNA-polyploid tumours were significantly more oftenimmunohistochemistry-positive than DNA paradiploid tumours(12/16 vs. 5/18, p 5 0.006). Likewise, polyploid tumours con-tained gene mutations significantly more often than paradiploidtumours (11/13vs.4/13,p 5 0.006). Of 17 patients with polyploidtumours, 8 had significant levels of p53-Ab in their serum,compared with only 2 out of 17 with paradiploid tumours( p 5 0.02).

DISCUSSION

In this study, agreement among the different techniques used forthe detection ofp53 alterations was high. Immunohistochemistryand ELISA (detection of the ‘‘mutant’’ conformation) providedevidence for the accumulation of p53 protein in 16 cases. Molecu-lar analysis was performed for 14 of these, and all presentedmutations within exons 5 to 8. On the other hand, no genemutations were detected in 10 out of 11 IHC- and ELISA-mutant-conformation-negative tumours tested. Interestingly, 9/16 of thesetumours had a paradiploid DNA content and it has been shown thatthese tumours do usually not undergo endo-reduplication (Remvi-kos et al., 1997). Likewise, tumours with strictly DNA-diploidindices (1.00–1.08) are mostly located in the proximal colon, oftenexhibit the RER (replication error) phenotype and contain a lowerrate of p53 gene mutation as compared with distal tumours (forreview, see Remvikoset al.,1997).

p53-Ab are detected in various types of cancer, and theirfrequency has been shown to be strictly proportional to theoccurrence of p53 mutations (Lubinet al., 1995a). These p53-Abreflect a humoral response that occurs early during tumourdevelopment (Lubinet al., 1995b). p53-protein epitopes recog-nized by these antibodies are localized in the amino- and carboxy-termini of the protein, similar to those found in animals immunized

with wild-type p53 protein (Legroset al.,1994; Wildet al.,1995).Hence, the immune response to p53 is directed against immuno-dominant epitopes unrelated to the mutational hot spots(Schlichtholz et al., 1992). Most patients with p53-Ab have ap53-gene mutation that leads to p53-protein accumulation. There-fore, several studies have attempted to determine whether specificp53-gene mutations are more immunogenic than others, withconflicting results (Davidoffet al.,1992; Preudhommeet al.,1994;Winter et al.,1992; Schlichtholzet al.,1994; Guineeet al.,1995;Wild et al.,1995; von Brevernet al.,1996). Mutations associatedwith p53-Ab have been found to reside prominently in exons 5 and6 in breast cancers (Winteret al., 1992), in exons 5 or 7 in lungcancers (Schlichtholzet al., 1992), or to affect the DNA bindingregion (i.e.,exons 5, 7 or 8) in esophageal cancers (von Brevernetal., 1996). Despite the small number of patients with p53-Ab in ourseries, it can be noted that 3/3 mutations in exons 5 and 6 and 3/4mutations in exon 8 were associated with p53-Ab in contrast toonly 3/9 mutations in exon 7.

It has also been suggested that rather than the mutation typeperse,p53-protein accumulation might be the common link leading top53-Ab production (Preudhommeet al., 1994), and that p53-protein accumulation in the absence ofp53-gene mutation couldlead to immunization (Lubinet al., 1995a). In our study, onepatient (PT23) presented with p53-Ab, while no evidence of ap53-gene mutation was found. Since immunohistochemical analy-sis showed heterogeneous p53 staining, it is possible that thistumour contained a mutation located outside the region explored byDGGE (exons 5 to 8), which occurs in about 10% of colorectalcancers (Soussiet al., 1994). It is also possible that the specimenused for DNA analysis was from an area containing less than 10%mutated cells, or that p53-protein accumulation in this tumour wasindependent of a gene mutation.

Interestingly, p53-Ab have also been found in patients having atumour that do not present evidence of p53-protein accumulation,because of a frameshift/stop/splice mutation or in the absence ofmutation (Wild et al., 1995). In our series, 2 patients (PT6 andPT30) had p53-Ab without evidence of mutant-p53-protein accumu-lation although there was evidence for a mutation in exon 8. Thepresence of p53-Ab in these patients might be due to modificationin p53-antigen processing or presentation to the immune systemwithout protein accumulation. Finally, the presence of an otheroccult p53-Ab eliciting cancer is unlikely in view of the patientfollow-up.

In conclusion, our study shows that analyzingp53alterations incolorectal cancer using different methods provides concordantresults in most cases. Methodologically, careful selection of tumourspecimens is required when comparing different methods for theassessement ofp53 status in this cancer. Individual discrepanciessupport the current view of the complexity and difficulties inpredicting p53 self-immunization. Nevertheless, our results sug-gest that the differentp53 mutations in colorectal cancer vary intheir induction of serum p53-Ab production.

ACKNOWLEDGEMENTS

We thank Dr. T. Soussi for his critical review of the manuscriptand Mrs. D. Roche for her editorial assistance.

REFERENCES

ANGELOPOULOU, K., STRATIS, M. and DIAMANDIS , E.P., Humoral immuneresponse against p53 protein in patients with colorectal carcinomas.Int. J.Cancer,70,46–51 (1997).DAVIDOFF, A.M., IGLEHART, J.D. and MARKS, J.R., Immune response to p53is dependant upon p53/HSP70 complexes in breast cancers.Proc. nat.Acad. Sci. (Wash.),89,3439–3442 (1992).GUINEE, D.G. JR and 15OTHERS, Gender comparisons in human lung cancer:analysis of p53 mutations, anti-p53 serum antibodies and C-erbB-2expression.Carcinogenesis,16,993–1002 (1995).

HALL , P.A. and LANE, D.P., p53 in tumour pathology: can we trustimmunohistochemistry?—revisited.J. Pathol.,172,1–4 (1994).

HAMELIN , R., LAURENT-PUIG, P., OLSCHWANG, S., JEGO, N., ASSELAIN, B.,REMVIKOS, Y., GIRODET, J., SALMON, R.J. and THOMAS, G., Association ofp53 mutations with short survival in colorectal cancer.Gastroenterology,106,42–48 (1994).

HAMMEL , P., BOISSIER, B., CHAUMETTE, M.T., PIEDBOIS, P., ROTMAN, N.,KOUYOUMDJIAN, J.C., LUBIN, R., DELCHIER, J.C. and SOUSSI, T., Detection

717p53 PROTEIN IN COLORECTAL CANCER

and monitoring of serum p53 antibodies in patients with colorectal cancer.Gut,40,356–361 (1997).HOUBIERS, J.G.A., VAN DER BURG, S.H., VAN DE WATERING, L.M.G.,TOLLENAAR, R.A.E.M., BRAND, A., VAN DE VELDE, C.J.H. and MELIEF,C.J.M., Antibodies against p53 are associated with poor prognosis ofcolorectal cancer.Brit. J. Cancer,72,637–641 (1995).LEGROS, Y., LAFON, C. and SOUSSI, T., Linear antigenic sites defined by theB-cell response to human p53 are localized predominantly in the amino andcarboxy termini of the protein.Oncogene,9, 2071–2076 (1994).LUBIN, R., SCHLICHTHOLZ, B., TEILLAUD , J.L., GARAY, E., BUSSEL, A., WILD,C.P. and SOUSSI, T., p53 antibodies in patients with various types of cancer:identification and characterization.Clin. Cancer Res.,1, 1463–1469(1995a).LUBIN, R., ZALCMAN , G., BOUCHET, L., TREDANEL, J., LEGROS, Y., CAZALS,D., HIRSCH, A. and SOUSSI, T., Serum p53 antibodies as early markers oflung cancer.Nature Med.,1, 701–702 (1995b).PREUDHOMME, C., LUBIN, R., LEPELLEY, P., VANRUMBEKE, M. and FENAUX,P., Detection of serum anti-p53 antibodies and their correlation withp53mutations in myelodysplastic syndromes and acute myeloid leukemia.Leukemia,8, 1589–1591 (1994).REMVIKOS, Y., MULERIS, M., SALMON, R.J. and DUTRILLAUX , B., Colorectalcarcinogenesis: from chromosomal evolution pathways to molecular patho-genesis.Cancer Genet. Cytogenet.,93,63–73 (1997).REMVIKOS, Y., TOMINAGA, O., HAMMEL , P., LAURENT-PUIG, P., SALMON, R.J.,DUTRILLAUX , B. and THOMAS, G., Increased p53 protein content ofcolorectal tumours correlates with poor survival.Brit. J. Cancer, 66,758–764 (1992).SCHLICHTHOLZ, B., LEGROS, Y., GILLET, D., GAILLARD , C., MARTY, M.,LANE, D., CALVO, F. and SOUSSI, T., The immune response to p53 in breast

cancer patients is directed against immunodominant epitopes unrelated tothe mutational hot spot.Cancer Res.,52,6380–6384 (1992).

SCHLICHTHOLZ, B., TREDANIEL, J., LUBIN, R., ZALCMAN , G., HIRSCH, A. andSOUSSI, T., Analyses of p53 antibodies in sera of patients with lungcarcinoma define immunodominant regions in the p53 protein.Brit. J.Cancer,69,809–816 (1994).

SOUSSI, T., LEGROS, Y., LUBIN, R., ORY, K. and SCHLICHTHOLZ, B.,Multifactorial analysis ofp53alteration in human cancer: a review.Int. J.Cancer,57,1–9 (1994).

STEPHEN, C.W., HELMINEN, P. and LANE, D.P., Characterisation of epitopeson human p53 using phage-displayed peptide libraries: insights intoantibody peptide interactions.J. mol. Biol.,248,58–78 (1995).

TOMINAGA, O., HAMMEL , P., HAMELIN , R., NAGAWA, H., MUTO, T. andREMVIKOS, Y., Quantification and characterization of total cellular p53protein in colorectal cancer.Cytometry,31,1–7 (1997).

VON BREVERN, M.C., HOLLSTEIN, M.C., CAWLEY, H.M., DE BENEDETTI,V.M.G., BENNETT, W.P., LIANG, L., HE, A.G., ZHU, S.M., TURSZ, T., JANIN,N. and TRIVERS, G.E., Circulating anti-p53 antibodies in esophageal cancerpatients are found predominantly in individuals with p53 core domainmutations in their tumors.Cancer Res.,56,4917–4921 (1996).

WILD, C.P., RIDANPAA, M., ANTTILA , S., LUBIN, R., SOUSSI, T., HUSGAFVEL-FURSIAINEN, K. and VANINO, H., p53 antibodies in the sera of lung-cancerpatients: comparison withp53 mutations in the tumour tissue.Int. J.Cancer,64,176–181 (1995).

WINTER, S.F., MINNA, J.D., JOHNSON, B.E., TAKAHASHI , T., GAZDAR, A.F. andCARBONE, D.P., Development of antibodies against p53 in lung cancerpatients appears to be dependent on the type ofp53mutation.Cancer Res.,52,4168–4174 (1992).

718 HAMMEL ET AL.