The Inflammatory Response WithinDukes’ B Colorectal Cancers: Implications forProgression of Micrometastases and Patient SurvivalJohn Murphy, M.B., Gerald C. O’Sullivan, M.Ch., F.R.C.S.I., Garry Lee, M.B., F.R.C.Path.,Michael Madden, M.D., F.R.C.P.I., Fergus Shanahan, M.D., F.R.C.P.I., J. Kevin Collins, Ph.D., andIan C. Talbot, M.D., F.R.C.Path.
Departments of Surgery, and Medicine, and Pathology, Mercy Hospital, and The Cork Cancer ResearchCentre, University College Cork, National University of Ireland, Cork, Ireland; and Academic Department ofPathology and ICRF Colorectal Cancer Unit, St. Mark’s Hospital, London, England
OBJECTIVES: The aim of this study was to determine therelationship between the inflammatory response in primarycolorectal carcinomas, patient outcome, and the fate of bonemarrow micrometastases.
METHODS: The populations studied were a) 155 consecutivepatients with Dukes’ B colorectal cancer and follow-up fora mean of 5 yr, from the Mercy Hospital, Cork, and b) 260consecutive patients with rectal carcinoma Dukes’ B andfollow-up for .10 yr, from St. Mark’s Hospital, London.The primary tumor was assessed for the Jass and “Crohn’s-like” lymphoid reactions. In 36 consecutive patients, bonemarrow aspirates were examined for micrometastases be-fore and$6 months postoperatively.
RESULTS: The relationship between prognosis and the in-flammatory reactivity in the tumors was similar in bothpopulations. In the Mercy group there were two deathsamong 40 patients who had coexistent Jass and Crohn’s-likeinfiltrates. This contrasted with 25 deaths among 58 patientsin whom both infiltrates were absent (p , 0.005). Resultswere intermediate in cases in which either type of inflam-matory reaction was present alone. In the St. Mark’s patientssimilar prognostic differences were sustained for up to 10yr. Bone marrow micrometastases were present in 12/36patients preoperatively and in 14/36 postoperatively. Sevenof 12 patients with preoperative micrometastases were neg-ative postoperatively, indicating clearance of tumor cells.Nine of 24 who tested negative preoperatively had micro-metastases postoperatively. The clearance and presence ofpostoperative micrometastases was related to the immuno-logical responses in the primary tumor.
Bone marrow micrometastases have been reported to be amarker of poor prognosis in breast (1), lung (2), and colo-rectal carcinomas (3). The mechanisms by which micro-metastatic tumor deposits confer an adverse prognosis re-mains unclear, and there is additional uncertainty regardingthe longevity and tumorigenic behavior of these micrometa-static cells. We have reported that the presence of postop-erative bone marrow micrometastases in samples taken 6months after surgical resection in patients with gastrointes-tinal tumors, including colorectal carcinomas seems to be anindicator of residual disease and poor prognosis (4). Wefurther suggested that clearance of micrometastases aftersurgery might be a good prognostic indicator. The mecha-nisms by which the micrometastatic cells are eliminatedfrom the marrow or persist postoperatively remain to bedetermined. Among potential causes, an immunoinflam-matory reaction seems likely.
The importance of an immune response to primary colo-rectal carcinomas has been well established (5–8) and hasbeen incorporated into one prognostic classification for rec-tal cancer (9). Immune reactivity to micrometastases may bean important mechanism governing the persistence or clear-ance of disseminated tumor cells, because for cell survivaland metastatic growth they must avoid destruction. The easydetection of micrometastases in bone marrow represents anopportunity to study metastatic processes and related issuesof residual disease and tumor dormancy. Unlike other sitessuch as liver and lung, which are also likely to harbormicrodeposits of tumor, samples are easily obtained preop-eratively and sequentially either during or after therapy.This accessibility enabled us to examine the relationshipbetween the inflammatory response in the primary tumorand the fate of micrometastases in the postoperativeperiod.
Patient Selection and Clinical StagingTwo groups of patients were studied. The first was a groupof 155 consecutive patients, who received curative resec-tions in Mercy Hospital, Cork, between 1986 and 1996 forDukes’ B nonsynchronous colorectal cancer, without con-current inflammatory bowel disease or familial polyposiscoli. In this study group, the final 36 patients had bonemarrow aspirates studied for micrometastases. Patients withDukes’ stage C disease were excluded, as they receivedpostoperative chemotherapy. The study population con-sisted of 79 men and 76 women with a median age of 67 yr(mean 65.9 yr, SD 10.7 yr, range 32–88 yr). Follow-up datawere obtained for all patients by direct contact through theoutpatient clinic.
Preoperative clinical staging included chest radiography,ultrasonography, and abdominal computed tomography(CT) scan. A full laparotomy, bimanual palpation of theliver, exploration of the peritoneum, and radical resection ofthe primary tumor including mesentery of the bowel wascompleted in each case. Diagnostic and clinical investiga-tions for staging of all patients was in accordance with theguidelines of the human ethics committee for clinical re-search, National University of Ireland, Cork. Informed con-sent for surgery and bone marrow aspiration was obtained inall cases.
Histological examination showed tumor extension be-yond the muscularis propria and absence of nodal metasta-ses. All patients assessed for micrometastases were followedin a dedicated clinic for a mean duration of 3 yr at the timeof writing.
Survival data were obtained on a second group compris-ing 260 consecutive stage B patients who underwent cura-tive resection for nonsynchronous stage B rectal cancerbetween 1960 and 1970 at St. Mark’s Hospital, London, inwhom the Jass immune inflammatory infiltrate and Crohn’s-like infiltrate to the primary tumor was assessed. Follow-updata were obtained for all patients, for a mean of 10 yr. Inthis population there were 167 men and 93 women, with amedian age of 62 yr (mean 62.5 yr, SD 10.75 yr, range of29–91 yr). The inclusion of this patient group permittedconfirmation of findings by different observers and assess-ment of putative prognostic influences of inflammatory re-actions within tumors that were sustainable for 10 yr.
Collection and Analysis of Bone Marrow SamplesA total of 36 consecutive Mercy Hospital patients withDukes’ stage B tumors had bone marrow aspirate samplescollected from both iliac crests under aseptic techniquesboth preoperatively and 6–12 months postoperatively. Anincision over the iliac crest was used to enable the marrowaspirate to be taken without contamination by skin epithe-lium. The preoperative marrow samples were obtained be-fore opening the abdomen and thus before operative manip-ulation of the tumor. The unfractionated marrow was fixed
by addition dropwise with gentle shaking into 70% ethanoland was stored at220°C until use for flow cytometryanalysis. From a separate aliquot of marrow, mononuclearcells were prepared by Ficoll-Hypaque density centrifuga-tion for analysis by alkaline phosphatase–antialkaline phos-phatase (APAAP) techniques (10). Where availability ofadequate marrow permitted, both techniques were used foranalysis, and micrometastases were diagnosed when cyto-keratin positive cells were detected by either technique.
The flow cytometry procedures have been described andvalidated elsewhere (11). In brief, bone marrow sampleswere washed three times with phosphate-buffered saline(PBS) at 4°C and were stained with fluorescein isothiocya-nate (FITC) conjugated monoclonal antibody directedagainst cytokeratin 18 (Sigma, Poole, Dorset, England). Theantibody was diluted 1:20 in PBS, and 10ml of this stockantibody was added to a 100-ml sample of marrow. Afterwashing again in PBS, staining for DNA content was com-pleted with propidium iodide containing ribonuclease. Sam-ples were analyzed (per 105 events) using a cell cycleanalysis program on an Epics Coulter Elite flow cytometer(Coulter, Luton, England). Only double staining cells werecounted as positive, thereby excluding nonspecific stainingor spuriously stained debris (Fig. 1). Positive marrow con-taining cytokeratin-positive HT29 and OC1 cells and neg-ative (normal bone marrow) controls were included in allanalyses (11). Threshold and background fluorescence werecontrolled for by using an irrelevant FITC monoclonal an-tibody of the same species and isotype as anticytokeratin-18.
Direct visual confirmation of the presence or absence ofmicrometastatic cells was also performed by histochemistry
Figure 1. Example of bone marrow micrometastases demonstratedby flow cytometry. Nucleated cells were identified by propidiumiodide stain (PMT3) and the micrometastases within this popula-tion by staining with cytokeratin 18 (PMT2).
3608 Murphy et al. AJG – Vol. 95, No. 12, 2000
with cytokeratin 18 using the APAAP technique (3, 10) oncytospun slides after Ficoll-Hypaque density centrifugation(10) (Fig. 2).
Assessment of the Primary TumorHematoxylin and eosin–stained slides from the primarytumor were reviewed without knowledge of the bone mar-row or clinical status to assess the Jass lymphocytic infiltrate(5, 12) and the Crohn’s-like lymphoid reaction as describedby Graham and Appelman (6). The Jass lymphocytic infil-trate was deemed to be present when there was a looseconnective tissue lamina infiltrated by lymphocytes cover-ing most of the deepest advancing tumor margin (5, 12)(Fig. 3A, B). The Crohn’s-like lymphoid reaction waspresent when three or more large lymphoid aggregates werepresent in two consecutive low-power fields (43 objective,103 eyepiece) at the periphery of the tumor and usuallylocated at the interface of the muscularis propria externa andthe serosa (6) (Fig. 3C). Tumor differentiation was deter-
Figure 2. Example of micrometastases identified by staining forcytokeratin 18, using an alkaline phosphatase–antialkaline phos-phatase technique.
Figure 3. (A) Low-powered photomicrograph showing an infiltrative tumor growth pattern without a Jass lymphoid infiltrate (hematoxylinand eosin, magnification340). (B) Medium-powered photomicrograph showing a prominent Jass lymphoid infiltrate (hematoxylin andeosin, magnification3200). (C) Low-powered photomicrograph showing a prominent “Crohn’s-like” lymphoid reaction, with a prominentJass lymphoid infiltrate to a primary Dukes’ B colorectal carcinoma (hematoxylin and eosin, magnification340).
3609AJG – December, 2000 Inflammation and Dukes’ B Colorectal Cancer
mined to be poor or “other” in accordance with the WorldHealth Organization (WHO) criteria (13). Tumor invasive-ness was assessed as mild, moderate, or severe by thecriteria of Dukes (14).
Data AnalysisSurvival data were stratified according to the immunoreac-tivity of tumors and analyzed using the methods of Kaplanand Meier with log rank analysis. Only deaths from recur-rent carcinoma were recorded as events. The relative con-tributions of patient age, gender, inflammatory reactivity,tumor differentiation, extramural venous invasion, and ex-tramural tumor spread on prognosis were determined byCox proportional hazard analysis (15) with backward elim-ination. Differences between categorical variables were ex-amined using Fisher’s exact test (16).
RESULTS
Mercy Hospital PatientsThe survival of the 155 Dukes’ B patients, with a medianage of 67 yr, was 73.5% at 5 yr. There was a clear relation-ship between patient outcome and the inflammatory reac-tivity in the tumor. In 40 patients (26% of the population)who had coexistent Jass and Crohn’s-like inflammatory
responses, there were two tumor recurrences (5-yr survival90.8%). In contrast, there were 25 cancer-related deaths in58 patients (37% of population) in whom the primary tumorwas negative for both inflammatory responses (5-yr survival52.95%,p , 0.005). Results were intermediate in patientsexhibiting a Jass (n5 37, five deaths, 5-yr survival 84.2%)or Crohn’s-like (n5 20, three deaths, 5-yr survival 84.3%)response alone (Fig. 4A). Using Cox proportional hazardanalysis, patient age, combined Jass and Crohn’s-like reac-tions, tumor differentiation, and degree of extramural tumorinvasion each had independent prognostic significance(Table 1).
In the latter 36 patients, bone marrow aspirates wereexamined pre- and postoperatively for micrometastases byeither flow cytometry or APAAP methods. The methodsemployed in analysis preoperatively were flow cytometry in32 cases, APAAP in 29, and, both methods in 25, andpostoperatively were flow cytometry in 24 cases, APAAP in29, and both methods in 17. Twelve patients were found tohave micrometastases on preoperative bone marrow aspi-rates (detected by flow cytometry in nine, by APAAP in six,and by both in three patients), and 14 patients were positiveon postoperative studies (detected by flow cytometry ineight, by APAAP in nine, and by both in three patients). Of
Figure 4. Dukes’ B patients divided into four groups according to the inflammatory response to their primary tumors: (I) Jass and“Crohn’s-like” reactions, both present; (II) Jass positive, Crohn’s-like negative; (III) Jass negative, Crohn’s-like positive; (IV) Jass andCrohn’s-like reactions, both negative. (A) Probability of survival in 155 Dukes’ B colorectal carcinomas from Mercy Hospital. (B)Probability of survival in 260 Dukes’ B rectal carcinomas from St. Mark’s hospital.
3610 Murphy et al. AJG – Vol. 95, No. 12, 2000
the 24 patients who tested negative for micrometastasespreoperatively, nine developed micrometastases postopera-tively, indicating ade novooccurrence rate of 37.5%. Incontrast, seven of 12 patients with preoperative microme-tastases were negative postoperatively, indicating a clear-ance rate of 58%. The relationship between the inflamma-tory reaction and the presence of micrometastases pre- andpostoperatively is shown in Tables 2 and 3. There was noassociation between the presence of micrometastases pre-operatively or postoperatively with patient age or gender,tumor differentiation, extramural venous invasion, and ex-tent of tumor spread outside the bowel wall (data notshown).
The clearance and the presence of postoperative micro-metastases was associated with both the Jass and Crohn’s-like lymphoid reactions to the primary tumor. Of sevenpatients showing clearance of bone marrow micrometasta-ses all had a Jass inflammatory reaction, whereas three of 14patients who showed micrometastases postoperatively had aJass inflammatory reaction (p 5 0.001). The Crohn’s-likeinflammatory reaction showed a similar relationship. ACrohn’s-like inflammatory reaction was present in five ofseven patients showing clearance of micrometastases, com-pared with only one of 14 patients who had postoperative
micrometastases (p 5 0.006). In nine patients in whom bothJass and Crohn’s-like reactions were present, seven showedclearance and no patient showedde novooccurrences, per-sistence, or cancer-related deaths. Conversely, of 15 patientswith absence of both Jass and Crohn’s-like responses, 10had postoperative micrometastases (three persistent, sevende novorecurrences, 0 clearance), and six of these patients(five with micrometastases) died of recurrent disease. In 12patients who showed only one type of inflammatory re-sponse, four had postoperative micrometastases (two per-sistent, twode novorecurrences, two clearances), and therewere two cancer-related deaths, both of which were negativefor micrometastases.
The mean length of current follow-up for these 36 pa-tients was 3 yr, and eight patients died of recurrent carci-noma. Preoperative bone marrow micrometastases were notprognostically significant, with only one death in 14 patientswith preoperative bone marrow micrometastases. However,of 14 patients with postoperative bone marrow micrometas-tases, five died of metastatic recurrence (36%), comparedwith three cancer deaths among 22 patients (14%) who werenegative for micrometastases. To date there have been nocancer deaths in seven patients who demonstrated postop-erative clearance of micrometastases (p , 0.05) or in 16
Table 1. Multivariate Analysis of Prognostic Variables in Colorectal Cancers in Two Different Hospitals*
Variable Estimated Coefficient SE Hazard Ratio (95% CI) p Value
3611AJG – December, 2000 Inflammation and Dukes’ B Colorectal Cancer
patients with a positive Jass inflammatory infiltrate (p ,0.01), including three patients with postoperative bone mar-row micrometastases. Of 14 patients with a Crohn’s-likeinflammatory infiltrate, two died. A single patient with aprominent inflammatory infiltrate who was negative forpostoperative bone marrow micrometastases died of a cere-brovascular accident without evidence of recurrence.
St. Mark’s PatientsFor the 260 Dukes’ B patients with a median age of 62 yr,survival was 81.2% at 5 yr and 75.8% at 10 yr. The prog-nostic significance of the Jass and Crohn’s-like infiltrate wasseparately confirmed by different observers (Fig. 4B), indi-cating that assessment of immunological responsiveness toprimary tumors is reproducible. In 35 patients (14% of thepopulation) who had both a Jass infiltrate and Crohn’s-likereaction, there were two deaths (5- and 10-yr survival, 97%and 93.7%, respectively). In contrast there were 30 deathsamong 124 patients in whom the primary tumors werenegative for both Jass and Crohn’s-like infiltrates, (5- and10-yr survival, 73.3% and 67.9%,p 5 0.006 by comparisonwith double positive group). Results were intermediate inpatients who demonstrated a Jass (n5 45, seven deaths, 5-and 10-yr survival, 85.7% and 83%) or Crohn’s-like (n556, 11 deaths, 5- and 10-yr survival, 82% and 77.3%)infiltrate alone (Fig. 4B). Similar to the Mercy populationpatient age, the combined Jass and Crohn’s-like reactionsand the degree of extramural invasion each had independentprognostic significance (Table 1).
DISCUSSION
The results of our study confirm the prognostic significanceof the inflammatory reaction to the primary tumor. In twoseparate populations, they demonstrate an interactive effectbetween the Jass and the Crohn’s-like inflammatory cellinfiltrates that confers a survival advantage. This disease-free survival advantage was sustained for up to 10 yr,suggesting cure by total ablation of the primary tumor andits metastatic cells, rather than deferment of tumor recur-rence. These findings have direct clinical relevance, as theyshow that histopathological assessment of the inflammatoryreaction in the primary tumor is a reproducible prognosticindicator. They further suggest the need for prospectivestudies to determine whether adjuvant therapy is indicatedin the subgroup of patients with Dukes’ stage B cancersshowing both Jass and Crohn’s-like inflammatory responsesin the tumor.
Although this study is retrospective in nature, completefollow-up data for cancer recurrence and cause of deathwere available through regular clinic attendance. There wasagreement between both groups of patients on the prognos-tic significance of age, immune reactivity of the tumor, andextramural extension of the cancer. Gender, tumor growthpattern, and venous invasion were independent prognosti-cators in the St. Mark’s patients but not in the Mercy
population, although tumor growth pattern and venous in-vasion had prognostic significance by univariate log-rankanalysis in the Mercy population. This may be due to minordifferences between both groups of patients in demograph-ics (the Mercy population was slightly older with an equalsex distribution, whereas in the St. Mark’s patients had amale preponderance) and the distribution of tumors (theMercy cancers were colorectal and the St. Mark’s were allrectal cancers). It is also possible that with longer follow-upand greater patient numbers, some of these latter variablesmay also assume independent significance within the Mercypatients.
In this series the presence of an inflammatory reaction, asassessed by both methods, was associated with clearance ofmicrometastases from the bone marrow. The absence of aninflammatory response was associated with postoperativebone marrow micrometastases and development of overtmetastatic disease. These results indicate that the inflamma-tory response is not only locally active but also has systemicantitumor effects. The survival to date of three patients witha Jass infiltrate and one patient with a Crohn’s-like reactionwho had postoperative micrometastases without progressionto overt metastases is consistent with previous findings thatcancer dormancy may depend on a functioning immunesystem (17, 18). It is likely that other organs such as the liverand lung may also harbor dormant deposits of residualdisease. A subsequent decline in immune activity by immu-nosuppresion, aging, or the emergence of tumor clonesresistant to immunological control may result in the failureof the immune system to inhibit the growth of these mi-crodeposits and may account for the late presentation ofdistant metastases in patients who have had potentiallycurative resections of a primary colorectal carcinoma.
This study differs from a previous report that found thepresence of preoperative bone marrow micrometastases tobe an independent predictor of survival in colorectal carci-nomas (3). This may be explained in part by sample size,disease stage, and demographic differences between studies.Most of the mortalities in the study of Lindemannet al. (3)occurred in patients with Dukes’ stage C disease. Morerecently, Dukes’ C patients receive adjuvant chemotherapyand are not therefore available for comparative studies preand postoperatively. In the same patient population withsimilar stage disease (Dukes’ B), we found the presence ofthe inflammatory cell infiltrate to be a superior predictor ofsurvival than the presence of micrometastases preopera-tively. By comparison of data obtained pre- and postoper-atively, we were able to identify variations in metastatic cellbehavior such as clearance, persistence, and dormancy. Theassociation between improved survival, inflammatory reac-tivity, and metastatic cell clearance or dormancy suggeststhe presence of a potent antitumor mechanism. The corol-lary supports this conclusion in that there is an associationbetween reduced survival prospects, absence of inflamma-tory reactivity, and presence of micrometastases in the post-operative aspirates.
3612 Murphy et al. AJG – Vol. 95, No. 12, 2000
The mechanisms that determine the inflammatory reac-tivity and associated tumor containment are uncertain. Itcould be argued that in this particular group the metastaticcells had a low or absent tumorigenic potential and weredestined to die and to prime the immune system to respondspecifically to the primary tumor. Pantelet al. have foundmicrometastases to express proliferative markers and togrow in culture (19). We have extended these studies inesophagogastric cancer and found these cells to be tumori-genic in athymic nude mice (20). Taken together, thesefindings suggest that the micrometastatic cells are viable,but further large-scale studies must be constructed to deter-mine whether there are intrinsic metastatic cell propertiesthat determine variables in tumorigenic behavior and hostresponses. The observation of three patients with post op-erative micrometastases who have been clinically tumor-free for .3 yr and who had a prominent immune infiltratein the primary cancer shows that viable metastatic cells cancoexist in a dormant state with antitumor responses. In theimmune-negative tumor group, there are four patients withbone marrow micrometastases postoperatively who have.3 yr of tumor-free survival. This suggests that metastaticcell dormancy may not, at least in all circumstances, beimmunologically determined. It is, however, possible thatthere may exist antitumor responses of varying efficacy thatcannot be predicted from study of the primary tumor. Mostlikely, dormancy is partly intrinsic to the metastatic cell andpartly immunogenic. The expression of genes for angio-genic growth factors and invasion of tissues are also pre-requisites for metastatic growth (21).
Notwithstanding the limitations of the present study, it isthe first attempt to address the relationship between tumorimmune reactivity and the fate of micrometastases in pa-tients who, by clinical and histological criteria, show noevidence of tumor dissemination. The smallness of the sam-ple size and the comparative insensitivity of iliac crestaspirates for micrometastases (20) preclude a more detailedmultivariate analysis of the relative contributions of meta-static cell behavior and tumor immune responsiveness tosurvival. Nevertheless, this study confirms the prognosticsignificance of the inflammatory reaction in the primarytumor and establishes a relationship with the fate of sys-temically disseminated micrometastases. Future studies willneed to address the genetic and molecular factors that de-termine clearance, dormancy, and progression of microme-tastases. Understanding the mechanisms of the intratumoralinflammatory reaction and its precise immune basis hasimportance in the development of effective systemic, im-mune-based or biological therapies in the future. Recentstudies suggest that tumor cells survive in a hostile immu-nological environment such as the lymph nodes and bonemarrow, by using a variety of strategies. These includefailure of antigen expression through absence of MHC mol-ecules in the tumor cell (22, 23), immunosuppression fromsoluble mediators elaborated by the cancer (24), resistanceby tumor cells to lysis by immune cells (25, 26) or direct
killing of Fas-sensitive lymphocytes by Fas ligand ex-pressed by Fas-resistant tumor cells—the Fas counter-attack(27, 28).
In summary, the association between the inflammatoryreaction and the clearance of micrometastases from bonemarrow indicates a potential systemic antitumor reactionthat confers a survival advantage. We propose that exami-nation of the primary tumor and investigations for micro-metastases pre- and postoperatively may lead to an in-creased understanding of how disseminated tumor cellsavoid destruction by the immune system and develop intoclinically relevant metastases. The results of such studiesmay assist in the development of effective immune or bio-logical therapies. Furthermore, our results suggest that pa-tients with Dukes’ stage B colorectal carcinoma whoseprimary tumors demonstrate both a Jass and Crohn’s typeimmune infiltrate in tumor have a probability of survival.90% at 5 yr. These results also suggest the need for largerprospective studies to establish whether adjuvant chemo-therapy has a role in this subgroup of patients.
ACKNOWLEDGMENTS
Supported by the Cancer Research Appeal Mercy Hospital(CRAM), the Imperial Cancer Research Fund, and in part bythe Health Research Board of Ireland. We appreciate theassistance of Marian McCarthy and the staff of the histo-pathology laboratory, Mercy Hospital, and gratefully ac-knowledge the technical assistance of Jackie Kelly, BernieCrowley, Jim O’ Callaghan, and Dr. Fiona O’ Brien. Wethank Dr. Jim Fitzgibbon, Department of Histopathology,Mercy Hospital, Cork, for helpful advice in the preparationof this study.
Reprint requests and correspondence:Professor G. C. O’ Sul-livan, Cork Cancer Research Centre, Research Laboratory, MercyHospital, Cork, Ireland.
Received Aug. 10, 1999; accepted July 24, 2000.
REFERENCES
1. Mansi JL, Berger U, Easton D, et al. Micrometastases in bonemarrow in patients with primary breast cancer: Evaluation asan early predictor of bone metastases. Br Med J 1987;295:1093–6.
2. Pantel K, Izbicki J, Passlick B, et al. Frequency and prognosticsignificance of isolated tumor cells in bone marrow of patientswith non-small-cell lung cancer without overt metastases.Lancet 1996;347:649–53.
3. Lindemann F, Schimok G, Dirschedl P, et al. Prognosticsignificance of micrometastatic tumor cells in bone marrow ofcolorectal patients. Lancet 1992;340:685–9.
4. O’ Sullivan GC, Collins JK, Kelly J, et al. Micrometastases:Marker of metastatic potential or evidence of residual disease.Gut 1997;40:512–5.
5. Jass JR. Lymphocytic infiltration and survival in rectal carci-noma. J Clin Pathol 1986;39:585–9.
3613AJG – December, 2000 Inflammation and Dukes’ B Colorectal Cancer
6. Graham DM, Appelman HD. Crohn’s-like lymphoid reactionand colorectal cancer: A potential histologic prognosticator.Mod Pathol 1990;3:332–5.
7. Ropponen KM, Eskelinen MJ, Lipponen PK, et al. Prognosticvalue of tumor infiltrating lymphocytes (TIL’s) in colorectalcancer. J Pathol 1997;182:318–24.
8. Harrison JC, Dean PJ, El-Zeky F, et al. From Dukes throughJass: Pathological prognostic indicators in rectal cancer. HumPathol 1994;25:498–505.
9. Jass JR, Love SB, Northover JMA. A new prognostic classi-fication of rectal cancer. Lancet 1987;1:1303–6.
10. Cordell JL, Falini B, Erber WN, et al. Immunoenzymaticlabelling of monoclonal antibodies using immune complexesof alkaline phosphatase and monoclonal antialkaline phospha-tase (APAAP complexes). J Histochem Cytochem 1984;32:219–29.
11. O’ Sullivan GC, Collins JK, O’ Brien F, et al. Micrometastasesin bone marrow of patients undergoing curative surgery forgastrointestinal cancer. Gastroenterology 1995;109:1535–40.
12. Jass JR, Ajioka Y, Allen JP, et al. Assessment of invasivegrowth pattern and lymphocytic infiltration in colorectal can-cer. Histopathology 1996;28:543–8.
13. Jass JR, Atkin WS, Cuzick J, et al. The grading of rectalcancer; historical perspective and multivariant analysis of 447cases. Histopathology 1986;10:437–59.
14. Dukes CE, Bussey HJR. Spread of rectal cancer and its effectson prognosis. B J Cancer 1958;12:1016–23.
15. Cox DR. Regression models and life tables. JR Stat Soc1972;34:187–220.
16. Owen F, Jones R. Statistics. London: Pitman Publishing, 1994;428–50.
17. Muller M, Gounari F, Prifti S, et al. EblacZ tumor dormancyin bone marrow and lymph nodes: Active control of prolifer-ating tumor cells by CD81 immune T cells. Cancer Res1998;58:5439–46.
18. Meltzer A. Dormancy and breast cancer. J Surg Oncol 1990;43:181–8.
19. Pantel K, Dickmanns A, Zippelius A, et al. Establishment ofmicrometastatic cell lines; a novel source of tumor cell vac-cines. J Natl Cancer Inst 1995;87:1162–8.
20. O’ Sullivan GC, Sheehan D, Clarke A, et al. Micrometastasesin esophagogastric cancer: High detection rate in resected ribsegments. Gastroenterology 1999;116:543–8.
21. Fidler IJ. Molecular biology of cancer: Invasion and metasta-ses. In: De Vita, Hellman, Rosenberg, eds. Cancer: Principlesand practice of oncology. Philadelphia and New York: Lip-pincott-Raven 1997:135–52.
22. Diederichsen AC, Ostenhom AC, Kronborg O, et al. Flow cyto-metric investigation of immune-response-related surface mole-cules on human colorectal cancers. Int J Cancer 1998;79:283–7.
23. Kaklamanis L, Townsend A, Doussis-Anagnostopoulou IA, etal. Loss of major histocompatibility complex-encoded trans-porter associated with antigen presentation (TAP) in colorectalcancer. Am J Pathol 1994;145:505–9.
24. O’ Sullivan GC, Corbett AR, Shanahan F, et al. Regionalimmunesuppression in esophageal squamous cancer: Evidencefrom functional studies with matched lymph nodes. J Immunol1996;157:4717–20.
25. Kammerer R, von Kleist S. CEA expression of colorectaladenocarcinomas is correlated with their resistance againstLAK-cell lysis. Int J Cancer 1994;57:341–7.
26. Zhang K, Sikut R, Hansson GC. A MUC1 mucin secretedfrom a colon carcinoma cell line inhibits target cell lysis bynatural killer cells. Cell Immunol 1997;176:158–65.
27. Alderson M, Tough RTW, Davis-Smith T, et al. Fas ligandmediates activation-induced cell death in human T-lympho-cytes. J Exp Med 1995;181:71–7.
28. O’ Connell J, O’ Sullivan GC, Collins JK, et al. The Fascounterattack: Fas mediated T cell killing by colon cancercells expressing Fas ligand. J Exp Med 1995;184:1075–82.
