In Situ Evidence of Neoplastic Cell Phagocytosis byMacrophages in Papillary Thyroid Cancer*

ANTONINO FIUMARA, ANTONINO BELFIORE, GIOVANNA RUSSO,EDVIGE SALOMONE, GABRIELLA M. SANTONOCITO, ORAZIO IPPOLITO,RICCARDO VIGNERI, AND PIETRO GANGEMI

Servizio di Anatomia ed Istologia Patologica, Presidio Ospedaliero Vittorio Emanuele II; and Istituto diMedicina Interna e di Malattie Endocrine e Metaboliche, Cattedra di Endocrinologia (A.B., G.M.S., R.V.),and Cattedra di Chirurgia (O.I., R.V.), University of Catania, Catania, Italy 95123

ABSTRACTWe evaluated the occurrence, tissue distribution, and prognostic

value of tumor-associated macrophages in 121 papillary thyroid car-cinomas using immunohistochemical staining with anti-CD68 anti-body in archival paraffin-embedded sections. Lymphocytic infiltra-tion and dendritic cell presence were also evaluated. Three groupswere identified according to the presence and characteristics of mac-rophages: 1) tumors without evidence of infiltrating macrophages:(n 5 35); 2) tumors with infiltrating macrophages but no evidence ofneoplastic cells phagocytosis (n 5 68); and 3) tumors with infiltratingmacrophages and in situ evidence of active neoplastic cell phagocy-tosis (n 5 18).

Neoplastic cell phagocytosis by macrophages was positively cor-related with both lymphocytic infiltration and dendritic cells (P 50.0000), whereas it was negatively correlated with vascular invasion(P 5 0.0032). Distant metastases developed in none of the 18 tumorswith neoplastic cell phagocytosis, but occurred in 15 of 103 of theremaining tumors (P 5 0.0647) and were significantly and negativelyassociated with lymphocytic infiltration or dendritic cells.

The present study indicates, therefore, that immune reaction, in-volving neoplastic cell phagocytosis by macrophages and lymphocyticinfiltration, plays a role in the development of distant metastases inpatients with papillary thyroid cancer. (J Clin Endocrinol Metab 82:1615–1620, 1997)

MACROPHAGES and lymphocytic infiltrates may befound in the context of or surrounding a variety of

malignancies and are commonly believed to represent thehost’s immune response to the tumor (1). According to thisview, a more favorable clinical evolution has been re-ported in various malignancies when associated withdense lymphocytic infiltrates (1). Less defined is the roleof tumor-associated macrophages, which represent a ma-jor component of the lymphoreticular infiltrates of tumors(2). As it is difficult to identify macrophages in conven-tionally stained tissue sections, few studies have ad-dressed the distribution and characteristics of tumor-as-sociated macrophages. The role of tumor-associatedmacrophages has, therefore, been studied mainly in in vivomodels (2).

Papillary cancer of the thyroid is a particularly suitablemodel to study tumor-associated macrophages, because it isfrequently associated with signs of immune reaction, includ-ing lymphocytic infiltration (3) and the presence of dendriticcells (4). Furthermore, molecules involved in the immunereaction, such as intercellular adhesion molecule and/orHLA class II molecules may be aberrantly expressed in trans-formed thyrocytes both in vivo and in vitro (5–7).

Therefore, we studied the tissue distribution and prog-

nostic significance of tumor-associated macrophages in aretrospective series of 121 papillary thyroid carcinomas.Macrophages were identified by immunohistochemistry,with the specific antibody anti-CD68, in paraffin-embeddedtissue sections (8, 9). We found tumor-infiltrating macro-phages in approximately 70% of the cases. Interestingly,phagocytosis of neoplastic cells by macrophages was ob-served in approximately 15% of tumors, and none of thesetumors developed distant metastases. These data suggestthat macrophages may affect papillary thyroid cancer pro-gression and, therefore, may be important targets for futureimmunotherapy.

Materials and MethodsTissues and patients

The study was carried out in archival, paraffin-embedded tissuesfrom 121 patients with thyroid papillary cancer. All patients had un-dergone total thyroidectomy plus pre- and paratracheal lymph nodedissection. Postsurgery, distant metastases were diagnosed when, inaddition to high serum thyroglobulin levels, the metastatic tissue waslocalized by total body scanning and, in most cases, by other imagingtests (standard x-rays, computed axial tomography scan, nuclear mag-netic resonance imaging, and bone scan). Diagnosis of regional lymphnode recurrence was made when ultrasound evidence of suspiciouslymph nodes was confirmed by either radioiodine uptake or a cytolog-ical finding of neoplastic epithelial cells and/or high thyroglobulinlevels in the aspirates.

Patients were classified for the extent of disease at presentation aspreviously described (10): class I, 41 patients with intrathyroid disease;class II, 24 patients with positive cervical lymph nodes; class III, 50patients with extrathyroid tumor invasion; and class IV, 6 patients withdistant metastases. The follow-up time of the 121 patients ranged from25–118 months (median, 63).

Received October 18, 1996. Revision received January 17, 1996.Accepted January 31, 1996.

Address all correspondence and requests for reprints to: AntoninoBelfiore, M.D., Istituto di Medicina Interna e di Malattie Endocrine e delMetabolismo, Cattedra di Endocrinologia, P.zza S. Maria di Gesu 1,95123 Catania, Italy. E-mail: endo.uni.ct@mail.tau.it.

* This work was supported in part by the Associazione Italiana perla Ricerca sul Cancro.

0021-972X/97/$03.00/0 Vol. 82, No. 5Journal of Clinical Endocrinology and Metabolism Printed in U.S.A.Copyright © 1997 by The Endocrine Society

1615

1616 FIUMARA ET AL. JCE & M • 1997Vol 82 • No 5

Histopathological examination

The diagnosis of papillary carcinoma was made according to theWHO criteria (11, 12). For each case all histological slides were reviewed,including multiple sections of all excised neoplastic and nonneoplasticthyroid tissue and lymph nodes. The occurrence of peritumoral lym-phocytic infiltrate and/or concomitant Hashimoto’s thyroiditis wasrecorded.

Immunohistochemical staining procedure

Monoclonal antibody anti-CD68, PGM1 clone, was used to identifymacrophages (9). It was used at a 1:100 dilution after treatment withtrypsin (0.5 mg/mL). An anti-S-100 protein polyclonal antibody (1:300dilution) was used to stain dendritic cells (4). Anti-epithelial membraneantigen (EMA) monoclonal antibody, E29 clone, was used (1:100 dilu-tion) as a marker of thyroid epithelial cells (13). Anti-CD20 and antiCD45-RO antibodies were used to identify B and T lymphocytes, re-spectively. All antibodies were obtained from Dako (Copenhagen,Denmark).

Paraffin-embedded sections were deparaffinized, incubated with ap-proximately 100 mL of the primary antibody at room temperature for 30min, and then processed for the biotin-streptavidin-peroxidase methodusing 3-amino-9-ethylcarbazole as substrate. Counterstaining was car-ried out with Mayer’s hematoxylin. Negative controls included omissionof the primary antibody. In selective cases a double immunohistochem-ical staining was carried out using as first layer the anti-CD68 antibodyfollowed by a biotinylated antibody and streptavidin-peroxidase plus3-amino-9-ethylcarbazole as revealing system. The anti-EMA antibodywas used as the second primary antibody, and the reaction was devel-oped by the alkaline phosphatase method using as substrate 5-bromo-4-cloro-3-indoxil phosphate/nitro blue tetrazolium chloride.

Statistical analysis

The distribution of clinicopathological characteristics was comparedin the different groups with the use of contingency tables and x2 test. Therelationship between metastasis-free interval and covariates as well ashazard ratios were calculated with the Cox proportional hazard regres-sion model (14). A Cox multivariate analysis, including all relevantprognostic factors indicated by univariate analysis, was carried outusing the occurrence of distant metastases as the end point. Kaplan-Meier plots were used to demonstrate the difference in the risk todevelop distant metastases between groups, and the log rank test usedto evaluate the differences between curves.

ResultsCharacteristics of tumor-infiltrating macrophages inpapillary thyroid tumors

According to results obtained by staining sections withanti-CD68 monoclonal antibody, the 121 papillary thyroidcarcinomas were subdivided into 3 groups.

Tumors without evidence of infiltrating macrophages. Thirty-fiveof 121 (28.9%) cases were included in this group. Low degreelymphocytic infiltrate was present in 5 cases. Dendritic cellswere present in 7 cases (20%). They were always located inthe papillar axis in close contact with the basal membraneand were often interdigitating the neoplastic cells lining thepapillary structures.

Tumors with infiltrating macrophages but no evidence of neoplasticcells phagocytosis. Sixty-eight of 121 (56.2%) cases were in-

cluded in this group. Either mononucleated macrophages orgiant multinucleated cells, foreign body type, were present.Accordingly, 2 subgroups were identified. The first group con-sisted of 44 of 121 (36.4%) tumors with mononucleated mac-rophages. CD68-positive cells could morphologically be distin-guished into 2 subtypes: a) small monocyte-like cells withintensely positive cytoplasm and small round nucleus; thesecells were located in the papillar axis or in the interfollicularstroma (Fig. 1A) and sometimes were seen in the process ofmigrating through the basal membrane; and b) medium/largemononucleated cells with a CD68-positive granule-rich cyto-plasm and a round or kidney-shaped nucleus; these cells werescattered in the papillar axis (especially medium size cells) orwere located in the interpapillar or in the follicular lumen (es-pecially large cells). In some cases, they were associated withregressive changes (hemorrhage and calcifications). In othercases, in the absence of regressive changes, large mono- orbinucleated cells were located in the interpapillar lumen in closecontact with neoplastic thyrocytes (Fig. 1B). In this group oftumors lymphocytic infiltrate was present in 25 of 44 (56.8%)cases, and dendritic cells were present in 19 of 44 (43.2%) cases.The second group consisted of 24 of 121 (19.8%) tumors withgiant multinucleated cells, foreign body type. In these cellsimmunostaining was mostly granular, sometimes diffuse, andless intense than that in mononucleated macrophages. Mono-layer sheets of epithelial neoplastic cells with regressive changespartially covered by CD68-positive coarse granules were some-times observed in the interpapillar lumen. Serial sectionsshowed that granule exocytosis from macrophages probablycaused this phenomenon. In this group, lymphocytic infiltratewas observed in 18 of 24 (75.0%) cases, and dendritic cells werepresent in 15 of 24 (62.5%) cases.

Tumors with infiltrating macrophages and in situ evidence ofactive neoplastic cell phagocytosis. Eighteen of 121 (14.9%) tu-mors were included in this group. These tumors were char-acterized by the presence of multinucleated CD68-positivecells. Large multinucleated (5–10 nuclei) cells with largeCD68-positive coarse cytoplasmic granules were present inthe interpapillar or intrafollicular lumen of these tumors (Fig.1, C–H). Some of them contacted the neoplastic cells withpseudopodia (Fig. 1D), some showing one or more phago-somes containing one or more neoplastic cells (Fig. 1, E–G).Neoplastic cells contacted by macrophages appeared of largesize, polygonal, intensely EMA positive, and detached fromadjacent cells (Fig. 1H). EMA positivity was gradually lost inthyroid neoplastic cells when they were included in a pha-gosome. Lymphocytes, either in contact with macrophagesor within a phagosome, were frequently observed. Mono-nucleated macrophages and giant multinucleated cells, for-eign body type, may also be present in these tumors. Lym-phocytic infiltration was observed in 15 cases (83.3%), anddendritic cell infiltration was seen in 17 cases (94.4%).

FIG. 1. Micrographs showing different phases of neoplastic cell phagocytosis by macrophages in papillary cancer of the thyroid. Immunohis-tochemical staining was carried out with anti-CD68 antibody in (A, B, D, E, and G), with anti-EMA antibody (H), and with the two antisera(CD68, red; EMA, black; E and F). Original magnification: A, 3200; B–H, 3400. A, Mononucleated macrophages scattered in the stroma; B,CD-68-positive mononucleated macrophages, in the process of fusing and sequestering neoplastic thyrocytes; C, D, F, H, multinucleatedmacrophages contacting neoplastic thyrocytes which shed into the interpapillary lumen; G and H, phagocytosis of neoplastic thyrocytes by one(G) or two multinucleated macrophages in the process of fusing (E).

THYROID CANCER AND MACROPHAGES 1617

Association with autoimmune thyroiditis

In 10 cases thyroid cancer was associated with histologicalevidence of diffuse lymphocytic thyroiditis. Nine of thesecases also had circulating antithyroid antibodies. All patientswere euthyroid, except one with subclinical hypothyroidism(TSH, 7.5 mol/L; normal, 0.3–4.5). Tumor-associated lym-phocytes (typically represented by T cells, some of themmigrating into the interpapillary or the intrafollicular luminaand in close contact with macrophages) were present in 8cases, dendritic cells in 5, and nonphagocytosing macro-phages in 8. At variance with tumor-associated lymphocytes,typical thyroiditis infiltrate was predominantly representedby B cells with scattered T cells and by germinal centers (anucleous of B cells surrounded by T cells).

Tumor characteristics at presentation and tumor outcomein relation to tumor-associated macrophages

The presence of macrophages and neoplastic cell phago-cytosis was positively associated with both lymphocytic in-filtration and the presence of dendritic cells (P 5 0.000) andwas negatively associated with vascular invasion (P 5

0.0032; Table 1). A more advanced disease (classes III and IV)was observed in tumors without macrophages compared tothose with neoplastic cell phagocytosis (group 1 vs. group 3,P 5 0.0425; Table 2).

To evaluate tumor outcome, we used as end point theoccurrence of distant metastases, an event strongly associ-ated with tumor-related deaths (15). The cumulative fre-quency of distant metastases was 6 of 35 (17.1%) in group 1,9 of 68 (13.2%) in group 2, and 0 of 18 in group 3 (P 5 0.0647,group 1 vs. group 3, by log rank test; Fig. 2A).

Distant metastases occurred more frequently in tumorswithout lymphocytic infiltration than in those with lympho-cytic infiltration (14 of 58 5 24.1% vs. 1 of 63 5 1.6%; P 50.0012, by log rank test; Fig. 2B). Also, distant metastasesoccurred in 13 of 63 (20.6%) tumors without dendritic cellsand in 2 of 58 (3.4%) with dendritic cells (P 5 0.0102, by logrank test; Fig. 2C). Other established risk factors (age, sex,tumor size, and extrathyroid tumor extension) were alsosignificant predictors of distant metastases at univariateanalysis (Table 3). No distant metastases occurred in thegroup with lymphocytic thyroiditis. The exclusion of this

TABLE 1. Association between tumor-infiltrating macrophages and neoplastic cell phagocytosis with clinical and pathological papillarythyroid cancer characteristics in patients at the time of primary surgery

Variable Group 1(n 5 35)

Group 2(n 5 68)

Group 3(n 5 18) P value

GenderFemale 30 (85.7) 58 (85.3) 15 (83.3)Male 5 (14.3) 10 (14.7) 3 (16.7) 0.9720

Age (yr)#45 24 (68.6) 44 (64.7) 9 (50.0).45 11 (31.4) 24 (35.3) 9 (50.0) 0.3968

Tumor diameter (cm)#1 14 (40.0) 12 (17.6) 4 (22.2)1.1–4 18 (51.4) 47 (69.1) 14 (77.8).4 3 (8.6) 9 (13.2) 0 0.0628

Histological gradeI 20 (58.8) 37 (57.8) 8 (44.4)II 12 (35.3) 24 (37.5) 9 (50.0)III 2 (5.9) 3 (4.7) 1 (5.6) 0.8614

Multifocal tumorNo 23 (65.7) 37 (54.4) 11 (61.1)Yes 12 (34.3) 31 (45.6) 7 (38.9) 0.5301

Bilateral tumorNo 27 (77.1) 50 (73.5) 13 (72.2)Yes 8 (22.9) 18 (26.5) 5 (27.8) 0.8494

Lymph node metastasesNo 18 (51.4) 37 (54.4) 10 (55.6)Yes 17 (48.6) 31 (45.6) 8 (44.4) 0.9555

ExtrathyroidNo 18 (51.4) 37 (55.2) 13 (72.2)Yes 17 (48.6) 30 (44.8) 5 (27.8) 0.3292

Vascular invasionNo 27 (77.1) 64 (95.5) 18 (100)Yes 8 (22.9) 3 (4.5) 0 0.0032

Distant metastases at surgeryNo 32 (91.4) 65 (95.6) 18 (100)Yes 3 (8.6) 3 (4.4) 0 (—) 0.3769

Lymphocytic infiltrationNo 30 (85.7) 25 (36.8) 3 (16.7)Yes 5 (14.3) 43 (63.2) 15 (83.3) 0.0000

Dendritic cellsNo 28 (80.0) 34 (50.0) 1 (5.6)Yes 7 (20.0) 34 (50.0) 17 (94.4) 0.0000

Group 1, Absence of macrophages; group 2, presence of nonphagocytosing macrophages; group 3, presence of neoplastic cell phagocytosis bymacrophages. Values are expressed as numbers of patients, with percentages in parentheses.

1618 FIUMARA ET AL. JCE & M • 1997Vol 82 • No 5

group from the statistical analysis did not appreciablychange the results.

At Cox multivariate analysis only three variables weresignificantly related to the risk of developing distant metas-

tases: lymphocytic infiltration (inversely related; P 5 0.0054)and extrathyroid extension and age (directly related; P 50.0255 and P 5 0.0128, respectively; Table 4).

Discussion

In the present study we demonstrate in situ phagocytosisof neoplastic cells by tumor-infiltrating macrophages in asubstantial proportion (15%) of papillary thyroid carcino-mas. In addition, another 56% of papillary cancers had in-filtrating macrophages but no evidence of neoplastic cellphagocytosis. No infiltrating macrophages were detected inthe remaining tumors studied. Neoplastic cell phagocytosiswas observed only in the interpapillary or the intrafollicularlumen, and only multinucleated macrophages were involvedin the phagocytosis process. Neoplastic cell phagocytosiswas strongly associated with the presence of dendritic cellsand lymphocytic infiltration, suggesting, therefore, that it ispart of the complex immunological host reaction to the can-cer. Tumor necrosis, in contrast, was never associated withlymphocytic infiltration, dendritic cells, or phagocytosingmultinucleated large macrophages.

Immune recognition and control of papillary cancer havebeen suggested (3, 16), and these are supported by both thefairly frequent finding of this tumor at the in situ stage whenautopsy screening is carried out (17) and the frequent asso-ciation of this slowly progressing tumor with the presence ofautoimmune thyroiditis (18). Papillary cancers have alsobeen shown to have a better cell-mediated immune responsecompared to controls and to follicular cancers. In particular,they have the highest proportion of activated monocytes andthe highest activity in the leukocyte adherence inhibitionassay (19).

In the last decade, tumor-associated macrophages havebeen studied in a variety of human malignancies using dif-

TABLE 2. Tumor stages in patients with papillary tumorssubdivided on the basis of the characteristics of infiltratingmacrophages

Tumorstage

Group 1(n 5 35)

Group 2(n 5 68)

Group 3(n 5 18)

Class I 7 (20) 26 (38.2) 8 (44.4)Class II 8 (22.9) 11 (16.2) 5 (27.8)Class III 17 (48.6) 28 (41.2) 5 (27.8)Class IV 3 (8.6) 3 (4.4) 0

Group 1, Absence of macrophages; group 2, presence of nonphago-cytosing macrophages; group 3, presence of neoplastic cell phagocy-tosis by macrophages. Values are expressed as the number of patients,with percentages in parentheses.

FIG. 2. Cumulative risk (Kaplan-Meyer plots) of distant metastasesin patients with papillary carcinomas according to macrophages (A),lymphocytic infiltration (B), and dendritic cells (C).

TABLE 3. Univariate analysis (Cox regression model) of patientswith papillary thyroid carcinoma according to clinicopathologicalvariables, using distant metastases as an end point

VariableCumulative risk of distant metastases

Hazardsratio

95% confidenceinterval

P value

Lymphocytic infiltrationNo 1.00Yes 0.06 0.01–0.46 0.006

Dendritic cellsNo 1.00Yes 0.15 0.03–0.68 0.0235

Vascular invasionNo 1.00Yes 9.29 3.29–26.14 0.0000

Extrathyroid extensionNo 1.00Yes 5.85 1.65–20.75 0.0062

Age (yr)#45 1.00.45 3.87 1.32–11.33 0.0135

Tumor diameter (cm),1 1.001–4 1.71 0.37–7.94 0.4891.4 5.99 1.09–32.73 0.0388

SexFemale 1.00Male 3.02 1.03–8.86 0.0432

THYROID CANCER AND MACROPHAGES 1619

ferent immunohistochemical markers (20, 21). To the best ofour knowledge, however, evidence of in situ neoplastic cellsphagocytosis by macrophages has been reported only as anoccasional finding (22–24). One possible explanation is thatsensitive techniques were not used. Another possible expla-nation is that neoplastic cell phagocytosis occurs with a dis-crete frequency in only some tumors. As pathological exam-ination reflects the tissue in a steady state at the moment ofsurgical resection, and phagocytosis is a dynamic and shortlasting process, it is possible that this phenomenon is over-looked unless it is present at a high rate. When neoplastic cellphagocytosis is minimal, only mononucleated or giant/for-eign body type cells may be observed.

The present observations suggest that the activation of themacrophage response may affect tumor behavior in papillarythyroid cancer. In our series, the presence of neoplastic cellphagocytosis by macrophages was associated with a lessadvanced tumor stage at presentation and with a trend fora reduced risk of distant metastases at follow-up (P 5 0.0647).Furthermore, lymphocytic and dendritic cell infiltrationwere both strongly associated with neoplastic cell phagocy-tosis by macrophages, and both were associated with a fa-vorable outcome, as previously reported (4, 25).

The mechanisms underlying the negative association be-tween neoplastic cell phagocytosis by macrophages and dis-tant metastases remains speculative. The negative correla-tion between phagocytosing macrophages and tumorvascular invasion, on the one hand, and the possible activityof phagocytosing macrophages in destroying micrometasta-ses, on the other hand, are both potential mechanisms (26).

In conclusion, our study confirms that the immune systemplays a major role in the clinical evolution of papillary thy-roid cancer (3, 4, 25, 27), and together with other biologicalmarkers (28), the evaluation of the immune response may berelevant to the management of this tumor. It also indicatesthat phagocytosing macrophages are an important compo-nent of this immune response. Furthermore, our study sug-gests possible therapeutic strategies for metastatic papillarythyroid cancer by enhancing the tumoricidal activity of mac-rophages (26).

References

1. Underwood JCE. 1974 Lymphoreticular infiltration in human tumours: prog-nostic and biological implications: a review. Br J Cancer. 30:538–548.

2. Mantovani A, Bottazzi B, Colotta F, Sozzani S, Ruco L. 1992 The origin andfunction of tumor-associated macrophages. Immunol Today. 13:265–270.

3. Baker J. 1995 The immune response to papillary thyroid cancer. J Clin Endo-crinol Metab. 80:3419–3420.

4. Schroeder S, Schwarz W, Rehpenning W, Loning T, Bocker W. 1988 Den-dritic/Langherans cells and prognosis in patients with papillary thyroid car-cinomas. Immunocytochemical study of 106 thyroid neoplasms correlated tofollow-up data. Am J Clin Pathol. 89:295–300.

5. Belfiore A. Mauerhoff T. Pujol-Borrell R, et al. 1991 De novo HLA class II andenhanced HLA class I molecule expression in SV-40 transfected human thyroidepithelial cells. J Autoimmun. 4:397–414.

6. Lucas-Martin A, Foz-Sala M, Todd I, Bottazzo GF, Pujol-Borrell R. 1988Occurrence of thyrocyte HLA class II expression in a wide variety of thyroiddisease: relationship with lymphocytic infiltration and thyroid autoantibodies.J Clin Endocrinol Metab. 66:367–75.

7. Tolosa E, Roura C, Marti E, Belfiore A, Pujol-Borrell R. 1992 Induction ofICAM-1 but not LFA-3 in thyroid follicular cells. J Autoimmun. 5:1019–1025.

8. Kaluza J. 1992 The monoclonal antibody (MAB) CD 68 allows the immuno-cytochemical identification of macrophages in primary and metastatic braintumors in paraffin embedded tissues. Folia Histochem Cytobiol. 30:125–128.

9. Falini B, Flenghi L, Pileri S, et al. 1993 PG-M1: a new monoclonal antibodydirected against a fixative-resistant epitope on the macrophage-restricted formof the CD68 molecule. Am J Pathol. 142:1359–1372.

10. De Groot LJ, Kaplan EL, McCormick M, Strauss FH. 1990 Natural history,treatment, and course of papillary thyroid carcinoma. J Clin Endocrinol Metab.71:414–424.

11. Heidinger C, Williams ED, Sobin LH. 1988 World Health Organization his-tological typing of thyroid tumors, 2nd ed. Berlin: Springer-Verlag; 9–11.

12. Rosai J. 1993 Papillary carcinoma. Monogr Pathol. 35:138–165.13. Pinkus GS, Kurtin PJ. 1985 Epithelial membrane antigen–a diagnostic dis-

criminant in surgical pathology. Hum Pathol. 74:620–624.14. Cox DR. 1972 Regression models and life-tables. J R Stat Soc B. 34:187–202.15. Mazzaferri EL. 1993 Thyroid carcinoma: papillary and follicular. In: Mazza-

ferri L, Samaan NA, eds. Endocrine tumors. Cambridge: Blackwell; 278–333.16. Boyd CK, Baker JR. 1996 The immunology of thyroid cancer. Endocrinol

Metab Clin North Am. 25:159–179.17. Bisi H, Fernandes VS, deCamargo RYA, Koch L, Abdo AH, deBrito T. 1989

The prevalence of unsuspected thyroid pathology in 300 sequential autopsieswith special references to the incidental carcinoma. Cancer. 64:1888–1893.

18. Haapala AM, Soppi E, Morsky P, Salmi J, Laine S, Mattila J. 1994 Thyroidantibodies in association with thyroid malignancy. Simultaneous occurrenceof thyroiditis and thyroid malignancy. APMIS. 102:390–394.

19. Juhasz F, Boros P, Szegedi G, et al. 1989 Immunogenetic and immunologicstudies of differentiated thyroid cancer. Cancer. 63:1318–1326.

20. Allen CA, Hogg N. 1987 Association of colorectal tumor epithelium expressingHLA-D/DR with CD8 positive T-cells and mononuclear phagocytes. CancerRes. 47:2919–2923.

21. van Ravenswaay Claasen HH, Kluin PM, Fleuren GJ. 1992 Tumor infiltratingcells in human cancer. On the possible role of CD161 macrophages in anti-tumor cytotoxicity. Lab Invest. 67:166–174.

22. Factor SM, Biempica L, Ratner I, Ahuja KK, Biempica S. 1977 Carcinoma ofthe breast with multinucleated reactive stromal giant cells. Virchows Arch.374:1–12.

23. Hashimoto H, Koga S, Watanabe H, Munetobe E. 1980 Undifferentiatedcarcinoma of the thyroid gland with osteoclast-like giant cells. Acta Pathol Jpn.30:323–324.

24. Gaffey MJ, Lack EE, Christ ML, Weiss LM. 1991 Anaplastic thyroid carcinomawith osteoclast-like giant cells. A clinicopathologic, immunohistochemical andultrastructural study. Am J Surg Pathol. 15:160–168.

25. Matsubayashi S, Kawai K, Matsumoto, et al. 1995 The correlation betweenpapillary thyroid carcinoma and lymphocytic infiltration in the thyroid gland.J Clin Endocrinol Metab. 80:3421–3424.

26. Fidler IJ. 1985 Macrophages and metastases-a biological approach to cancertherapy: presidential address. Cancer Res. 45:4714–4726.

27. Belfiore A, Garofalo MR, Giuffrida D, et al. 1990 Increased aggressivenessof thyroid cancer in patients with Graves’ disease. J Clin Endocrinol Metab.70:830–835.

28. Farid NR, Shi Y, Zou M. 1994 Molecular basis of thyroid cancer. Endocr Rev.15:202–232.

TABLE 4. Multivariate analysis (Cox regression model) ofpatients with papillary thyroid carcinoma according to clinical andpathological variables, using distant metastases as an end point

VariableCumulative risk of distant metastases

Hazardsratio

95% confidenceinterval

P value

Lymphocytic infiltrationNo 1.00Yes 0.06 0.01–0.43 0.0054

Extrathyroid extensionNo 1.00Yes 4.35 1.20–15.80 0.0255

Age (yr)#45 1.00.45 3.97 1.19–15.80 0.0128

1620 FIUMARA ET AL. JCE & M • 1997Vol 82 • No 5