@

Daniel Kirgan, MD; Pacita Manalo, MD; Mark Hall, PhD; Byron McGregor, MD

PATIENTS AND METHODS

Tissue for study was obtained from the Anatomical PathologySection of the Veterans Affairs Medical Center, Reno, Nev. Thirtynormal colons, 30 single tubulovillous adenomas, and 43 carcinomas(30 invasive carcinomas and 13 carcinomas in situ) were selected.Archival paraffin-embedded biopsy specimens of these tissues werestudied. Immunohistochemical techniques were used to screen thetissue for fue presence of HPV antigen. Specimens testing positivefor the antigen underwent DNA hybridization. Four-micrometersections of the tissue were cut and mounted on slides prepared withpoly-L-lysine. Tissues were then deparaffiJÚzed in xylene and rehy-drated in a series of alcohol washes. Endogenous peroxidase activity\\'88 blocked with 3% hydrogen peroxide in methanol. After rinsing indistilled water, the specimens were covered ~th 5% normal goatserum in phosphate-buffered saline (PBS) and incubated for 20 min-utes. AIl incubation was carried out at room temperature in a closedhumid chamber. The goat serum \\'88 decanted and rabbit antiboVinepapillomavirus antibody (1:100 111 dilution in lo/c albumin in rES) wasapplied and incubated for 18 to 20 hours. The specimens were nextwashed in rES, and a biotinylated horse antirabbit IgG (1:100 ¡LLdilution) was applied and incubated for 60 minutes. Tissue was againwashed in rES; then avidin-biotin complex (Vector Inc, Burlingame,CaJif) was applied to the sections and incubated for 60 minutes.Following washes in rES, tissues were stained ~th 3,3, diaminoben-zidine tetrahydrochloride for 10 minutes and counterstained in hema-toxylin for 1 to 3 minutes. Tissues were dehydrated \\;th increasingconcentrations of ethanol, cleared with xylene, and mounted. Resul-tant slides were interpreted blindly by a pathologist and gradedsimply as positive or negative.

Specimens that reacted positive by immunohistochemistry weresubjected to in situ DNA hybridization. Four-micrometer sections oftissue were deparaffiJÚzed \\;th xylene and rehydrated using a seriesof alcohol washes. Tissues were then subjected to protease digestionusing a commercially prepared digestive reagent (Life Technologies,Gaithersburg, Md) for 20 minutes at 37.C, and then washed withbuffered saline solutions prior to DNA hybridization. Specimenswere reacted \\;th pooled DNA probes to HPV \'irot)'pes 6,11,16, 18,31, 33, and 35 (L~Tech) at 100.C for 5 minutes, then hybridizationwas allowed to occur at 37.C for 2 hours. Specimens were againwashed in buffered saline solutions (Life Tech) for 3 minutes andcounterstained. Tissues were then dehydrated in a series of increas-ing concentrations of alcohol, washed in xylene, and mounted onslides for photomicrography.

.Human paplllomavirus has been shown to be assoclatedwith squamous carcinomas. We evaluated benign and malignantcolon tissues 10r the presence 01 human papillomavirus in1ectionto determine 11 a similar relationship exlsts between human papil-lomavirus and colon neoplasms. Colon tissues were screenedusing an immunohistochemical technique to detect human papil-lomavlrus antigen. In situ DNA hybridization was then performedon those tissues that yielded positive results by immunohisto-chemistry. Groups were compared using x' analysls. Humanpapillomavlrus antigen was present in 23% 01 normal colon spec-imens, 60% 01 benign tumors, and 97% 01 carcinomas. Humanpapilloma viral geno me was demonstrated in 27% 01 benigntumors and In nearly 43% 01 all carcinomas tested. These dataindlcatethat human papillomavirus In1ects the columnar mucosa01 the colon, and that an association exists between humanpaplllomavirus and colon neoplasia.

(Arch Surg.1990;125:862-865)

T he incidence of adenocarcinoma of the colon and rectumhas remained unchanged for the last 40 years. Despite

perioperative improvements in nutrition, antibiotics, and an-esthesia, the mortality associated with colon carcinoma con-tinues to be significant; approximately hall of the 150 000newly diagnosed patients with colorectal cancer this year wiIldie of their disease. The causes of this neoplasm remainunknown. Despite associations of viral infections with otherneoplasms for over three quarters of a century, a viral caus-ative agent for colon cancer has not been proposed. SquamousceIl carcinomas of the urogenital region and aerodigestivetract have previously been associated with human papilloma-virus (HPV) infection and these relationships have provenclinicaIly useful in both the diagnosis and monitoring ofthesediseases. Demonstration of a similar relationship in coloncancer may have the same effect on this disease.

We have recently demonstrated an association betweenHPV antigen and both benign and malignant neoplasms ofthecolon.'" These studies, however, only demonstrated virus-related proteins and not the viral genome. In situ DNAhybridization is a quick and reproducible test for those viralgenomes and thus is specific for vira! infections.

We chose to screen colon tissues, including normal mucosa,benign tumors, and cancers for HPV infection using an im-munohistochemical technique to detect HPV antigen. In situDNA hybridization was then utilized to demonstrate HPVgenome in those colonic tissues that tested positive for viral

antigen.

RESULTS

Human papillomavirus antigen was demonstrated in 97% oíthe carcinomas (Fig 1). The antigen was seen in 60% oí thebenign tumors (Fig 2) and 23% oí the normal specimens (Fig3). Despite small numbers (n = 103), the null hypothesis thatthese subpopulations were homogenous in their incidence oívira! antigen was able to be rejected at P<.OO5 by Xl analysis.The viral genome was demonstrated in roughly one third oíthe carcinoma specimens tested (Fig 4), and two thirds oícarcinomas in situ reacted positive íor the virus. A smaller

Accepted íor pubJication April18, 1990.From the Departments oí Surgery (Drs McGregor and Kirgan), Pathology

(Dr Manalo), and Microbiology (Dr Hall), University oí Nevada School oíMedicine, Reno; and the Veterans Affairs Medical Center, Reno.

Reprint requests to Department oí Surgery, VA Medical Center, 1(KK)Locust, Reno, NV 89502 (Dr McGregor).

Human Papillomavirus & Colon Neoplasms-Kirgan el al862 Arch Surg-VoI125. July 1990

~

.'

Fig 1. -Left, Photomicrograph 01 human colon carcinoma demonstrating human papillomavirus antigen by an immunohis-tochemical technique. The diaminobenzidine tetrahydrochloride stains brown in the presence 01 antigen-antibody complex(hematoxylin, original magnification x 100). Right, Photomicrograph of human colon carcinoma that tested negative forhuman papillomavirus antigen. Reaction in the interstitium is due to nonspecific peroxidase within plasma cells (originalmagnification x 100).

Fig 2.-Left, Photomicrograph demonstrating human papillomavirus antigen in a benign tubulovillous adenoma (originalmagnification x 100). Right, Photomicrograph of ahuman tubulovillous adenoma testing negative for human papilloma-virus antigen (original magnification x 100).

I '~::;~

'<

;~:

~

# ;

.~

Fig 3.-Photomicrograph of normal human colon mucosa demon-strating human papillomavirus antigen (original magnification x 100).

Human Papillomavirus & Colon Neoplasms -Kirgan et al 863Arch Surg-VoI125, July 1990

,. ~ '~

*HPV indicates human papillomavirus.tP<.OO1 by X' compared wilh normal mucosa.

fraction ofthe benign tumors and none ofthe normal mucosasamples tested positive for the viral genome (Table).

COMMENT

The carcinogenic potential of the papilloma..irus has beenpreviously demonstrated,S and progression from a benignlaryngeal papilloma to carcinoma has been described.4 Themechanism by which oncogenesis occurs is not known, butretroviruses and DNA viruses are known to actívate proto-oncogenes.6 This is probably a multistep process with di-etary6.1 and genetic interactions.s,e Evidence is accumulatingthat supports a causal role for HPV in benign and malignantlesions ofthe urogenital tract and the head and neck regían.Carcinoma ofthe cervix,lo-1S aerodigestive tract,I4-11Iung,lS andesophagusle,~ have all been shown to have an association withHPV. Identification of this viral association has allowed forchanges in diagnostic and therapeutic approaches to thesediseases.

Proto-oncogene abnormalities have algo been identified incolon neoplasia, including the c-RAS and c-.~YC oncogenes.21Products ofthe c-MYC oncogene have been demonstrated byimmunoperoxidase-staining techniques in benign colonic pol-yps and carcinomas of the colon. Hybridization techniqueshave demonstrated the c-MYC oncogene in as many as 20% ofcolon cancers.22 Human papillomavirus is belie\'e4 to incorpo-rate into the host genome near the c-.~YC oncogene, at leastin carcinomas ofthe uterine cervix.2S Evidence that oncogenesare active in colon neoplasia adds strength to the argument ofmultistage carcinogenesis and is consistent \\'Íth the well-acknowledged progression from mucosa to adenomatouschange to malignant degeneration (the polyp-cancersequence).

We have previously demonstrated HPV antigen to be asso-ciated with colonic neoplasia. This association led us to inves-tigate whether the virus was actually present in colon tumorsor whether Borne other tumor antigen might have cross-reacted with the polyclonal antibody used in the immunohis-tochemical procedure. Hybridization of DNA remains thestandard technique for the demonstration and thus identifica-tion of viral infections, but conventional extraction DNAhybridization is cumbersome and time consuming. In situDNA hybridization has recently been described for detectionof viral genome,24 and was chosen for this study based on itsprevious success in similar situations. It has been shown to bea clinically useful technique in detecting HPV genome in

laryngeal papilloma tissue25 and cytomegalovirus in colon tis-sue ofimmunosuppressed individuals.~

The immunohistochemical demonstration of viral capsidprotein in one of four normal mucosal specimens suggests awidespread prevalence ofthe virus. Extrapolated to the esti-mated US population of 250 million, a 23% prevalence wouldimply that nearly 60 million American colons harbor the virus.Our tissue samples were from a geriatric population, and wehave no data on how or \\'hen viral infestation occurs, but sucha large number is consistent with the emerging knowledge ofthis virus.

The detection ofHPV genome in nearly half ofthe cancerstested clearly demonstrates the presence ofthe virus in thesecarcinomas. This validates the immunohistochemical datathat detected viral capsid antigen in these lesions, and coIlab-orates previous reports that identified the viral antigen incolon polyps. It is not clear why aIl of those tissues thatdisplayed the viral capsid protein did not alBo demonstrate theviral genome; either the HPV associated with those tumors isone of the other 53 known virotypes not represented in ourpooled probes or the virus is present in such low concentrationthat it is undetectable by in situ hybridization.

We have demonstrated the presence of HPV in the colum-Dar mucosa of the colon. Further, we have described anincreasing association of the virus with progressive degreesofmalignant degeneration ofthat mucosa. The potential ther-apeutic and diagnostic implications ofthis association remainto be established, but we believe that this association v.ithHPV wiIl have an impact on future understanding ofboth thedevelopment of colon carcinoma and its diagnostic and thera-peutic interventions.

References

localization ofpapilloma\irus antigens in cen'ical dysplasia and vuJ\'ar condylo-mas. Am J Obsret GyMcol. 1981;140:931-935.

11. Lancaster WD, Jenson AB. E\'idence for papillomavirus genus-specificantigens and DNA in,.t~'ngeal papilloma. Inrervirology. 1981;15:204-212.

12. Durst M, Gissl.,ann 1, Ikenberg H, zur Hausen H. A papilloma\irusDNA from a cen'ical carcinoma and its pre\'81ence in cancer biopsy samplesfrom different geographic regions. Proc Natl Acad Sci USA 1983;80:3812-3815.

13. Crum CP, Ikenberg H, Richart RM, Gissman L. Human papilloma\irustype 16 and early cen'icaI neoplasia. N E1/Dl J Med. 1985;310:880-883.

14. Kashima H, Wu T-C, Mounts P, Heffner D, Cachay A, Hyams V.Carcinoma ex-papilloma: histologic and \'irologic studies in whole-organ sec-tions ofthe larynx. Lary1/Doscope. 1988;98:619-624.

15. Corbitt G, Zarod AP, Arrand JR, Longson M, Farrington WT. Humanpapilloma\irus (HPV) genotypes associated with iaryngeal papilloma. J Glinfuthol. 1988;41:284-288.

16. BrandEma JL, Steinberg BM, Abramson AL, Winkler B. Presence ofhuman papillomavirus type 16 related sequences in verrucous carcinoma ofthelarynx. Gancer Re.. 1986;46:2185-2188.

17. Gissmann 1, Dielh V, Schultz-Coulon HJ, zur Hausen H. Molecularcloning and characterization of human papilloma\irus DNA derived from aiaryngeaI papilloma. J Virol. 1982;44:393-400.

18. Syrganen K, Syrganen S, Kellokoski J, Karja J, Mantyjarvi R. Humanpapilloma\-irus (HPV) type 6 and 16 DNA sequences in bronchial squamous cellcarcinoma demonstrated by in situ DNA hybridization. Lu1/D. 1989;167:33-42.

1. Kirgan D, Manalo P, McGregor B. Imrnunohistochemical demonstrationof human papilloma\"Írus antigen in human colon neoplasms. J Surg Res.

1990;48;397-402.2. Kirgan D, Manalo P, McGregor B. Distribution ofhuman papilloma\"Írus

in colon neoplasia. Presented at the annual meeting of the International SocietyofSurgery, September 12, 1989, Taranto, Canada.

3. Rous P, Kidd JG. The carcinogenic effect of a papilloma\iros on the tarred

skin ofrabbits. J Exp Med. 1938;67:399-428.4. Zarod AP, Rutherford JD, Corbitt G. Malignant pragression ofl~-ngeal

papilloma associated with human papilloma\iros tl"pe 6 (HPV-6) DNA. J Glinfuthnl. 1988;41:280-2&'3.

5. Land H, Parada L, Weinberg R. Cellular oncogenes and multistep carci-

nogenesis. Science.19&'3;18:771-777.6. Reddy BA, Wynder EL. Large-bowel carcinogenesis: fecal constituents

ofpopulations with diverse incidence rates of colon cancer. J .Vatl Gancer Inst.

1973;50:1437-1442.7. Burkitt DP. Etiology and prevention of colorectal cancer. Hosp Prar-t Off:

1984;19:67-77.8. Cannon-Albright LA, Skolnick MH, Bishop DT, Lee RG, Burt RW.

Camrnon inheritance ofsusceptibility to colonic adenomatous poll"Ps and associ-ated colorectal cancers. N Engl J Med. 1988;319:533-537.

9. Kune GA, Kune S, Watson LF. The raJe ofheredity in the etiology oflargebowel cancer: data fram the Melbourne coJorectal cancer study. World J Surg.

1989;13:124-131.10. Kurman RJ, Shah KH, Lancaster WD, Jenson AB. Imrnunoperoxidase

Human Papillomavirus & Colon Neoplasms- Kirgan el al864 Arch Surg-VoI125. July 1990

19. Wmkler B, Capo V, Reumann W, et al. Human papilloma\irus infectionafilie esophagus. CanA:el: 1985;55:149-155.

20. KulskiJ, DemeterT, SterrettG, Shilkin K. HumanpapillomavirusDNAin oesophageal carcinoma. LanA:et. 1986;2:683.

21. Ste~oart J, E\'Bn G, Watson J, Sikora K. Detection of the c-MYC onca-gene product in colonic polyps and carcinomas. Br J Cancer. 1986;53:1-6.

22. Meltzer SJ, Ahnen OJ, Battifora H, Yokota J, Cline M. Protooncogeneabnormalities in colon cancers and adenomatous pol~'Ps. Gastroenterol;¡gy.1987;92:1174-1180.

23. Broker T, Thomas R. Structure and genetic expression oí papilloma\ir-

Discu

MARVIN M. ROMSDAHL, MD, PhD, Houston, Tex: Dr Kirgan andbis associares have examined benign and malignant colonic tissue forHPV and have provided evidence that probably should be inrerpreredas preliminary in regard to an association between HPV and coloncancer. Clearly the issue oí colon cancer causation by HPV is neithermade nor implied.

Human papillomavirus has previously been rather strictly associ-ared with squamous epithelium, either in the oral pharynx or theanorectal regions. Demonstration oí HPV in glandular epitheliumwould be a new observation. Ifindeed this should prove to be the case,and particularly ü associated with glandular-type carcinoma, a majarscientific finding would be at bando A confirmed association to coloncancer, even "ithout reíerence to causation, would be extremelyimportant.

Thia type oí study will receive extreme scrutiny in regard torechnology employed and its inrerpretations because of the inconsis-rency oí these findings "ith generalIy accepted concepts oí coloncarcinogenesis.

The authors have previously shown virus-associared proteins incolonic tissues, including benign and malignant neoplasms, by histo-chemical rechniques. They then proceeded to employ DN A hybridiza-tion to derect whether the virus was contained in the nucleus oí thedifferent colon tissues. WhiIe the immunohistochemical evidence isreasonably evident, a more exact and specific test íor HPV concernsthe DNA hybridization, which they indeed have done. Now, in regard.to DNA hybridization, 1 would like to point out that there are severalrechniques to show .this particular phenomenon.

The first rechnique and the oldest is the in situ labeling, usingtissuesections, such as they have done. A second rechnique that has beendeveloped is rermed the "dot blot" technique. For this rest, homoge-llares oí the tissue that you wish to examine, such as rectal mucosa,colon cancer, or polyps are alIowed to react with your indicator DN A,which would, oí course, come írom any oí the virus subtypes. This,then, can be tesred either immunochemically or by radioimmuno-assay.

The final t~'Pe oí rest íor DNA hybridization concerns what isknown as the "southern blot" technique. This is a rest in which ahomogenare afilie rest tissue is submitted.to gel electrophoresis andthen radioirnmunoassay is used to detect whether indeed there hasbeen hybridization between the papillomavirus and the DNA con-tained in the colonic tissues. Oí these different rechniques, the south-ern blot rechnique is the uItimare standard and should be employed bythe authors to document their findings at this time.

The authors did test seven different HPV viral types, but underlow stringency. It would be worthwhile.to rest these same viral t~'Pesunder high stringency. Perhaps in their closing remarks they \\iIl

Human Papillomavirus & Colon Neoplasms -Kirgan et al 865Arch Surg-VoI125. July 1990

uses. Obstet GyMcol Glin North Am. 1987;14:329-348.24. Brigati D, Myerson D, Leary J, et al. Detection of viral genomes in

cultured cells and paraffin-embedded tissue sections using biotin-labeled hy-bridization probes. Virotogy. 1983;126:32-50.

25. TerryR, LewisF, GriffithsS, WellsM, BirdC. Demonstrationofhumanpapilloma\irus types 6 and 11 in juvenile laryngeal papillómatosis by in situDNA hybridization. J futhol. 1987;153:245-248.

26. Roberts W, Hammond S, Sneddon J, Thesing J, Caldwell J, Clausen K.In situ DNA hybridization for cytomegalo\irus in colonoscopic biopsies. Archfuthol Lab Med. 1988;112:1106-1109.

ssion

discuss stringency for this audience and possible advantages thathigh stringency might have in adding credibility to their scientificobservations.

The observations that Dr Kirgan and his associates have made mayhave significant implications in colon cancer, should an undisputedassociation with HPV be forthcoming. However, this is a revolution-ary concept and I believe that use of the highest scientific technologyhas to be employed to support their findings.

Finally, it is a pleasure to see general surgeons and surgical train-ees familiarizing themselves with and engaging in basic research ofconditions we regularly encounter, such as colon cancer.

DAVIDL. BOUWMAN, MD, Detroit, Mich: Do you have any cases inwhich, in a single individual, assays were ron on normal colonicmucosa as well as the villous adenoma, tubular adenoma, orcarcinoma?

GEORGE B. MARKLE IV, MD, Carlsbad, NM: I have long beenimpressed with the simiJarities between colon cancer and carcinomaof the cervix: dysplasia first, then going into carcinoma in situ, andfinally frank carcinoma. So there's a similarity here. Where does this

-virus come from ifwe are dealing with a papillomavirus? In gynecolo-gy we know where it comes from and we can do something about it.Are there any ideas as to how this virus gets into the gut?

DR MCGREGOR: We have many ofthe same concems expressed byDr Romsdahl and other reviewers. 'lb be sure, gene amplification andsouthemblot techniques are the state of the art, even though thosefiltration techniques would not have allowed the simultaneous identi-fication and localization as did the in situ hybridization.

Dr Bouwman, we had no cases in which all three were involved.Second, all ofthe "normal" mucosa d.escribed was taken from patientswith nonmalignant diseases. Dr Markle, we have speculated aboutthe origin of this virus and have completed a distribution study oftubulovillous adenomas with the theory that if there were a gradientfrom rectum to cecum, a retrograde transmission from the perineumwould be suggested, but ifthe distribution were homogenous, then aviremic presentation would be more likely. The result of that studywas that viral distribution was uniform throughout the colon.

But at the basis of all the skepticism, especially among surgeons ist.his: Do we believe all this fancy biochemistry? The validation of anysuch observation is that it allows us to predict future observations.One ofthose predictions surelywould be thatifitis there, we ought tobe able to seeit. Indeed, we have recently been able to demonstrate:(1) central clearing of chromatin, (2) openings in the nuclear envelopecharacteristic of a lytic virus, and (3) 45 to 55 nm viral particles (thesize ofHPV) all within the nucleus of a cell of a human colon adenocar-cinoma. We believe that this is a real phenomenon and is worthy offurther study.