Malignant Lymphoma of the Thyroid Gland a Clinicopathologic Study of 108 Cases

Malignant Lymphoma of the Thyroid Gland A Clinicopathologic Study of 108 Cases

Gregory A. Derringer, M.D., Lester D. R. Thompson, M.D., R. Allen Frommelt, Ph.D., Karen E. Bijwaard, M.S., Clara S. Heffess, M.D., COL, MC, USA, and Susan L. Abbondanzo, M.D.

We report a retrospective clinicopathologic study of 108 pri- with appropriate therapy, but prognosis depends on both clini- mary thyroid gland lymphomas (PTLs), classified using the cal stage and histology. MZBL and stage IE tumors have an REAL and proposed WHO classification schemes. The patients excellent prognosis, whereas tumors with a large cell compo- included 79 women and 29 men, with an average age of 64.3 nent or DLBCL or stage greater than IE have the greatest years. All patients presented with a thyroid mass. The PTLs potential for a poor outcome. were classified as marginal zone B-cell lymphoma of mucosa- Key words: Thyroid—Malignant lymphoma—Classification dif- —Marginal zone B-cell lymphoma—Mucosa-associated ,(30 ס associated lymphoid tissue (MALT) or MZBL (n —lymphoid tissue (MALT)—Immunohistochemistry ,(36 ס fuse large B-cell lymphoma (DLBCL) with MZBL (n .and follicle center lym- Molecular analysis—Prognosis ,(41 ס DLBCL without MZBL (n -Excluding the FCL, features of lympho .(1 ס phoma (FCL; n mas of MALT-type were identified in all groups, despite a Am J Surg Pathol 24(5): 623–639, 2000. follicular architecture in 23% of cases. Lymphocytic thyroiditis (LT) was identified in 94%. Ninety-one percent of patients presented with stage IE or IIE disease, whereas 69% had perithyroidal soft tissue infiltration. All patients were treated with Primary thyroid gland lymphomas (PTLs) are uncom- surgical excision followed by adjuvant therapy (76%): chemo- mon tumors that have been estimated to represent ap- therapy (15%), radiation (19%), or a combination of radiation proximately 5% of all thyroid neoplasms73 and 2.5% to and chemotherapy (42%). Disease-specific survival was 82% at 7% of all extranodal lymphomas.10,24,59 Studies have last follow up (mean, 82.8 mos) and 79% at 5 years. Statisti- shown that PTLs typically occur in middle- to older-aged cally, stages greater than IE, presence of DLBCL, rapid clinical growth, abundant apoptosis, presence of vascular invasion, women and arise in the setting of autoimmune thyroiditis high mitotic rate, and infiltration of the perithyroidal soft tissue (lymphocytic thyroiditis, Hashimoto’s disease). The rela- were significantly associated with death with disease. No pa- tive risk of a patient with lymphocytic thyroiditis devel- tients with MZBL or stage IE disease died with disease. In oping lymphoma has been estimated to be 40 to 80 times summary, PTLs typically occur in middle- to older-aged indi- greater than in the general population29,43,59 and to take, viduals as a thyroid mass, with a predilection for females. De- spite their histologic heterogeneity and frequent simulation of on average, 20 to 30 years to develop after the onset of 59 other lymphoma subtypes, virtually all PTLs are lymphomas of lymphocytic thyroiditis. MALT-type arising in the setting of LT. Mixed DLBCL and In the past, PTLs have been classified using a variety MZBL are common. Overall, PTLs have a favorable outcome of different terms because of the use of older histologic classifications, implying variable clinical behavior.3,5,9,13,19,23,24,30,41,42,45,48,51,52,54,63,64,70,71,74,75,77,80 From the Departments of Hematopathology (G.A.D., S.L.A.), Endocrine and Otorhinolaryngic-Head & Neck Pathology (L.D.R.T., There is a better understanding of the concept of mucosa- C.S.H.), Cellular Pathology (K.E.B.), and Department of Epidemiol- associated lymphoid tissue (MALT) and recognition of ogy, Repository and Research Services (R.A.F.), Armed Forces the lymphomas that can arise from MALT because of the Institute of Pathology, Washington, DC, U.S.A. 35 Presented at the 87th Annual Meeting of the United States and initial description by Isaacson and Wright. In light of Canadian Academy of athology, Boston, MA, February 28–March 7, the MALT concept, certain authors think PTLs represent 1998. a single clinicopathologic entity: MALT lymphoma.3,36 The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the In the Revised European-American Classification of views of the Department of the Army or the Department of Defense. Lymphoid Neoplasms (REAL)27 and in the proposed Address correspondence and reprint requests to Gregory A. World Health Organization (WHO) Classification for Derringer, MD, Department of Hematopathology, Building 54, Room 40 G124a, Armed Forces Institute of Pathology, 6825 16th Street, NW, Neoplastic Diseases of the Lymphoid Tissues, these Washington, DC 20306-6000, U.S.A. distinct lymphoid tumors are included under the term

623 624 G. A. DERRINGER ET AL.

“marginal zone B-cell lymphoma of MALT-type and extension into the perithyroidal adipose tissue and/or (MZBL).” skeletal muscle. Fibrosis was graded semiquantitatively Whereas large cell transformation of MZBL (mixed as mild (thin bands or small microscopic foci of fibrosis DLBCL and MZBL) is recognized and has been reported on high-power magnification [40× objective]), moderate in the thyroid34,45,59,69 and studied in other extranodal (broad bands of fibrosis on intermediate magnification sites,17,22,30,46,60,61,66 its characterization, frequency, and [20× objective]), and severe (large areas of sclerosis re- effect on clinical behavior remains unclear in the thyroid placing the thyroid architecture on low-power magnifi- gland. Furthermore, there are no large, long-term clini- cation [10× objective]). Mitotic activity was rated as low copathologic studies of PTLs using the MALT concept (less than 10 mitotic figures/10 HPF) or high (greater and the most recent lymphoma classification schemes than 10 mitotic figures/10 HPF), using the 40× objective. (REAL and proposed WHO). Therefore, we undertook a Diffuse large cell lymphoma was defined as a lymphoma retrospective study of 108 PTLs using current concepts or areas within a small cell lymphoma that were com- to characterize the clinical, histomorphologic, immuno- prised of large lymphoid cells that “sheeted out” in a phenotypic, and molecular characteristics of PTLs. diffuse manner. Increased, scattered, single, large immu- noblasts or Reed-Sternberg-like cells in a small cell lym- MATERIALS AND METHODS phoma or a predominance of large atypical cells in the follicles of a small cell lymphoma were not considered One hundred eight cases of PTL, seen in consultation indicative of a large cell lymphoma. All cases were in- from 1985 to 1993, were retrieved from the files of the vestigated for the presence of lymphocytic thyroiditis, Endocrine Registry at the Armed Forces Institute of defined in this study as reactive lymphocytic or lympho- Pathology, Washington, DC. They were selected from a plasmacytic infiltrates in the thyroid parenchyma with total of 170 cases (2.4% of 7083 benign and malignant formation of lymphoid aggregates and fibrosis. The pres- primary thyroid gland neoplasms seen in consultation ence of follicles with germinal center formation and oxy- during the same period). The authors defined PTLs as philic change of the follicular epithelium was not re- lymphomas that presented primarily as thyroid gland tu- quired for this designation. Other thyroid abnormalities mors. Cases were excluded if there was inadequate clini- were also documented. cal history and follow up, a lack of hematoxylin and With the aid of Musshoff’s modification of the Ann eosin (H&E)-stained sections, insufficient material for Arbor staging system,16 PTLs were staged as follows: immunohistochemical analysis, or a history of lym- IE: PTLs with or without extension into the perithyroidal phoma. Ninety-four cases were obtained from civilian sources, including university medical centers and foreign soft tissue; IIE: PTLs with involvement of lymph nodes contributors, 11 cases from military hospitals, and three on the same side of the diaphragm; IIIE: PTLs with cases from Veterans Administration Medical Centers. involvement of lymph nodes on both sides of the dia- Materials were supplemented by a review of surgical phragm and/or spleen; and IVE: PTLs with dissemina- pathology and operative reports, cancer registry records, tion to other extranodal sites. and written questionnaires or oral communication with Formalin-fixed, paraffin-embedded tissue and un- the treating physician(s). Follow-up information in- stained sections were available for H&E and periodic cluded patients’ demographics, symptoms at presenta- acid Schiff (PAS) stains and immunohistochemical tion, radiographic studies, clinical and surgical staging of analysis. Immunophenotypic analysis was performed ac- the lymphoma, treatment modalities, and the current sta- cording to the standardized avidin–biotin method of Hsu 32 tus of the disease and patient. et al. using a pertinent commercially available antibody The lymphomas were classified with the aid of estab- panel (Table 1). Predigestion was performed with either lished criteria, recent studies, and the REAL and pro- protease or pepsin (as noted): 3 minutes with 0.05% posed WHO classifications.27,31,40 They were evaluated Protease VIII (Sigma Chemical Co, St. Louis, MO, for features of MALT or lymphomas of MALT-type, USA) in a 0.1 M phosphate buffer at pH of 7.8 at 37°C including lymphoepithelial lesions (LEL), monocytoid or 20 minutes with 0.04% pepsin in a 0.1 M phosphate B-cells, centrocyte-like cells, plasma cells, abnormal cy- buffer at pH 2.0 at 40°C. Antigen enhancement (recov- toplasmic immunoglobulin accumulations, Dutcher bod- ery) was performed when appropriate by using formalin- ies, and reactive follicles with or without colonization by fixed, paraffin-embedded tissue treated with a buffered lymphoma (“follicular colonization”). Additional fea- citric acid solution and heated for 20 minutes in a cali- tures assessed were presence or absence of abundant ap- brated microwave oven. Following this, the sections optosis (more than four apoptotic cells per high-power were allowed to cool at room temperature in a citric acid field [HPF] using the 40× objective), tumor necrosis, buffer solution for 15 to 45 minutes before continuing vascular invasion (infiltration of neoplastic cells into the the procedure. Standard positive and negative (serum) wall of any thyroidal vessel [artery, vein, or lymphatic]), controls were used throughout.

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 625

TABLE 1. Immunohistochemical panel

Antibody Primary antibody Company Dilution Antigen unmasking

CD3 Rabbit polyclonal Dako Corporation, 1:500 Protease treatment Carpinteria, CA CD5 Mouse monoclonal Vector/Novocastra Labs, 1:100 Microwave pretreatment Burlingame, CA CD10 Mouse monoclonal Vector/Novocastra 1:100 Microwave pretreatment CD20 (L26) Mouse monoclonal Dako 1:200 — CD23 Mouse monoclonal The Binding Site, 1:200 Pepsin treatment San Diego, CA CD43 (MT1) Mouse monoclonal Biotest, Denville, NJ 1:50 — CD45RA (MT2) Mouse monoclonal Biotest 1:20 — CD45RB (LCA) Mouse monoclonal Dako 1:200 — CD45RO (UCHL-1) Mouse monoclonal Dako 1:200 — CD79a Mouse monoclonal Dako 1:200 Microwave pretreatment Bcl-1 (cyclin D1) Mouse monoclonal Immunotech, Marseilles, France 1:3000 Microwave pretreatment Bcl-2 Mouse monoclonal Dako 1:100 Microwave pretreatment Beta F-1 Mouse monoclonal Endogen, Woburn, MA 1:50 Pepsin treatment Kappa Rabbit polyclonal Dako 1:25000 Protease treatment Lambda Rabbit polyclonal Dako 1:50000 Protease treatment Cytokeratin Mouse monoclonal Boehringer Mannheim, 1:50 & 1:200 Protease treatment (AE1/AE3 and CK1) Indianapolis, IN Thyroglobulin Mouse monoclonal Dako 1:100 —

Polymerase Chain Reaction needed in each group to have adequate power to detect statistically meaningful differences. DNA was extracted from formalin-fixed, paraffin- 65 For the purpose of this summary, strong findings were embedded sections as previously described. t(11;14) defined as those results for which the p value was less and t(14;18) bcl-2/MBR-JH PCR assays were performed 2,81 than 0.05 and the confidence interval did not include the and analyzed as previously described. The t(14;18) null value. Weak findings were when either the p value bcl -2/MCR-JH assay was performed as the MBR-JH as- or the confidence interval was statistically significant, say using AmpliTaq Gold DNA polymerase for the “hot- but not both. Stages IIIE and IVE were combined in the Bcl-2 primers were derived start” application. /MCR-JH analysis because of the limits of sample size. from published sequences.55,65 t(14;18) positive control cell lines (SUDHL-6 and SUDHL-16) were obtained from Alan Epstein, MD, PhD (USC, Los Angeles, CA, RESULTS USA). Pathology Statistical Analysis Macroscopic Findings Statistical analysis was performed on 107 cases, excluding the single case of FCL. Mean differences be- The lymphomas ranged in size from 0.5 to 19.5 cm in tween groups were compared using unpaired t tests or maximum dimension with a mean size of 6.9 cm (Table analysis of variance (ANOVA), respectively. Tukey 2) without a statistically significant difference in tumor multiple comparison procedures were used to make two- size between the three groups. Tumors that presented as way comparisons among groups in which ANOVA was a rapidly enlarging mass were, on average, larger (7.2 used. Differences between observed and expected cm) than those tumors that presented as a mass only (6.4 .(0.644 ס frequencies were compared using chi-square tests or cm), but this was not statistically significant (p or ,(23 ס left (n ,(33 ס Fisher’s exact test when there were fewer than five pa- The tumors involved the right (n of the thyroid gland. The specific (48 ס tients in any subcategory. When necessary, the cases both lobes (n were divided into two main groups, MZBL and DLBCL, lobe of the thyroid gland involved was unknown in four or lymphomas with a large cell component, which in- cases. cluded both mixed DLBCL and MZBL and DLBCL The tumors were described grossly as firm to soft, without MZBL groups. lobulated, multinodular to diffuse, solid or cystic masses. Survival analysis and a survival comparison between The cut surface was smooth or slightly bulging, pale, tan the two main groups of lymphomas, MZBL and DLBCL, to white–gray, or red with a “fish-flesh,” uniform, ho- could not be performed because only 20 patients died mogeneous to mottled appearance. Foci of hemorrhage with disease (all with DLBCL). At least 30 cases are and necrosis were frequently noted. Attachment to the

Am J Surg Pathol, Vol. 24, No. 5, 2000 626 G. A. DERRINGER ET AL.

TABLE 2. Macroscopic and microscopic features of the main histologic categories of PTLs

Mixed DLBCL DLBCL without MZBL and MZBL MZBL Pathologic feature N=30(no.) N=36(no.) (%, no.)

Maximum size of tumor 6.8 cm 7.7 cm 6.1 cm Lymphocytic thyroiditis 100% (30) 97% (35) 85% (35) Perithyroidal extension 53% (16) 64% (23) 85% (35) Vascular invasion 0% (0) 19% (7) 49% (20) Lymphoepithelial lesion 100% (30) 97% (35) 70% (29) “MALT ball”-type LEL 67% (20) 61% (22) 10% (4) Marginal zone/monocytoid B-cell differentiation 97% (29) 83% (30) 32% (13) Plasmacytoid differentiation 100% (30) 92% (33) 17% (7) Plasma cell inclusions 47% (14) 42% (15) 0% (0) Reactive follicles 90% (27) 78% (28) 17% (7) Colonized follicles 87% (26) 75% (27) 12% (5) Fibrosis Mild 43% (13) 25% (9) 42% (17) Moderate 37% (11) 44% (16) 32% (13) Severe 20% (6) 31% (11) 27% (11) Mitotic rate Low 57% (17) 3% (1) 5% (2) High 43% (13) 97% (35) 95% (39) Necrosis 10% (3) 69% (25) 73% (30) Abundant apoptosis 13% (4) 86% (31) 93% (38) Adenomatoid nodules 20% (6) 22% (8) 20% (8) Microscopic papillary carcinoma 10% (3) 6% (2) 5% (2)

PTLs, primary thyroid gland lymphomas; LEL, lymphoepithelial lesions; MZBL, marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue type; DLBCL, diffuse large B-cell lymphoma. surrounding adipose tissue or skeletal muscle was de- licular epithelium (Fig. 3). The second type was more scribed in 75 cases. unique, characterized by variably sized, rounded masses of neoplastic cells that filled and distended the colloidal space of the follicles, termed by us as “MALT ball” LEL Microscopic Findings (Fig. 3). ס The lymphomas were classified into four groups: 30 Lymphocytic thyroiditis was present in 94% (n MZBL, 36 mixed DLBCL and MZBL, 41 DLBCL with- 101), including the case of FCL. Lymphocytic thyroiditis out MZBL, and one FCL. All lymphomas were com- usually contained reactive follicles with an absence of a prised of B-lymphocytes that effaced the thyroid paren- dense, sheet-like effacement of the thyroid parenchyma chyma (Fig. 1). Although the lymphomas varied archi- and demonstrated both B- and T-cells areas by immuno- tecturally and cytologically from case to case and within histochemistry. Neoplastic cells occasionally colonized an individual case, features of lymphomas of MALT- the germinal centers of reactive follicles in areas of lym- type were identified in almost all cases. phocytic thyroiditis. The uninvolved thyroid parenchyma Infiltration by the lymphoma into the perithyroidal contained adenomatoid nodules which were occasionally -microscopic papil ,(22 ס adipose tissue and/or skeletal muscle was detected in calcified and degenerated (n and ,(1 ס follicular adenoma (n ,(7 ס of 107 cases assessable), including the case of lary carcinoma (n 75) 70% .(1 ס FCL. The frequency of soft tissue infiltration increased follicular carcinoma (n as the quantity of the large cell component increased Marginal Zone B-Cell Lymphoma of MALT- -with 58 of these 75 patients having a Type (MZBL) (n = 30). MZBL consisted of predomi ,(0.027 ס p) DLBCL. Twenty-seven cases (25%) demonstrated vas- nantly small lymphoid cells with variable proportions of centrocyte-like cells, plasma cells, lymphoplasmacytoid ס cular invasion (Fig. 2) and all were DLBCL (p 0.0001). A variable degree of background fibrosis was lymphocytes, monocytoid B-cells, and interspersed large detected in all cases. transformed lymphocytes (Fig. 4). LEL were identified Although quantity and quality varied from case to in all cases. In a single case, a rare large transformed cell case, lymphoepithelial lesions (LEL) could be identified with Reed-Sternberg-like morphology was noted, but in most cases. LEL occurred in two forms. The first form DLBCL was not identified. The monocytoid B-cells, fre- was more common and of the usual type, characterized quently arranged in cohesive groups, varied from small by neoplastic cells in variably sized nests or rounded to intermediate-sized cells containing round, oval, or groupings infiltrating within and between the thyroid fol- slightly irregular nuclei with dark chromatin and moder-

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 627

ture was so extensive in five cases that it recapitulated a follicle center lymphoma (Fig. 6). A single case had the appearance of a mantle cell lymphoma (MCL) because it was nodular, was comprised of predominantly small irregular lymphocytes, and had a significant mitotic rate. Correct classification was made in these cases by iden- tifying the characteristic histologic features of MZBL aided by immunohistochemical and molecular analysis. Mixed DLBCL and MZBL (Marginal Zone B-Cell Lymphoma of MALT-Type With Large Cell Trans- formation) (n = 36). The MZBL portion had similar histologic features to those described already. The DLBCL was associated with either a transitional zone of increasing numbers of large cells intermixed with smaller neoplastic lymphocytes or occurred as an abrupt proliferation of large lymphoid cells surrounded by typical MZBL (Fig. 7). Variably present in each case, the large cells demonstrated centroblastic-like, im- munoblastic, monocytoid B-cell, and plasmacytoid differentiation (Fig. 8). Scattered Reed-Sternberg-like im- munoblastic cells were seen in two cases. At least focally, the areas of DLBCL in nine cases had Burkitt-like morphology consisting of sheets of intermediate-sized hyperchromatic, round neoplastic lymphocytes with mild nuclear irregularities and small amounts of basophilic cytoplasm, associated with a brisk mitotic activity, sig-

FIG. 1. Lymphocytic thyroiditis with germinal centers (top). Malignant lymphoma effaces the thyroid parenchyma (bottom). ate to abundant amounts of clear or pale eosinophilic cytoplasm. Cytologic atypia in monocytoid B-cells was occasionally noted and recognized by nuclear enlargement with open chromatin and conspicuous nucleoli. The monocytoid B-cells appeared epithelioid in some cases and, in others, were prominent, simulating, on low- power magnification, a follicle center lymphoma or a small lymphocytic lymphoma with large growth (proliferation) centers. Plasma cells were recognized in all cases, but plasmacytoid differentiation was prominent, at least focally, in 13 cases. An almost pure plasma cell population was seen in two cases. Dutcher bodies and cytoplasmic immunoglobulin (Ig) accumulations (crystalline Ig and Mott cells) were identified in 47% of cases, particularly impressive in three (Fig. 5). Reactive secondary lymphoid follicles, with or without follicular colonization and/or a follicular pattern, were found in 27 (90%) cases. In 11, a significant follicular architecture pattern was seen, largely attributed to follicular colonization. However, the follicular architec- FIG. 2. Vascular invasion by malignant lymphoma.

Am J Surg Pathol, Vol. 24, No. 5, 2000 628 G. A. DERRINGER ET AL.

FIG. 3. (A) A DLBCL infiltrates the thyroid follicular epithelium, but not the colloid space, forming classic LEL. (B) CD20 highlights neoplastic B-cells (left) and cytokeratin accentuates residual thyroid follicular epithelium (right). (Same case as A.) (C) Neoplastic cells of MZBL filling the colloidal space and expanding it, creating a “ball-like” appearance, termed “MALT ball” LEL. nificant apoptosis, and a “starry-sky” pattern on low- power magnification. The DLBCL component ranged from small, single or multiple foci to greater than 90% of the area of the tumor. Usually, the DLBCL component represented more than 50% of the entire lymphoma. LEL were identified in nearly every case of the mixed DLBCL and MZBL group, albeit more obvious in the MZBL areas. In the areas of DLBCL, the LEL were usually smaller, more widely separated from one another, and tended to be of the usual type (described previously). Reactive secondary follicles with or without colonization by neoplastic cells were found in 78% of mixed DLBCL and MZBL. A FIG. 4. A MZBL with a mixture of lymphocytes, mono- follicular architecture was noteworthy in 14 cases. In cytoid B-cells, plasma cells, and rare transformed lympho- three lymphomas, the follicular architecture was particu- cytes, forming LEL.

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 629

FIG. 5. Abnormal immunoglobulin accumulation as crystalline cytoplasmic inclusions. larly prominent and, in conjunction with the DLBCL areas, simulated a follicle center lymphoma, large cell with follicular and diffuse areas. However, these cases were correctly classified by the identification of features of lymphomas of MALT-type and immunohistochemical and molecular analysis. Diffuse Large B-Cell Lymphoma Without MZBL (DLBCL Without MZBL) (n = 41). The DLBCLs were large cell lymphomas that lacked an associated MZBL. The large neoplastic cells of this group demonstrated the

FIG. 7. (A) Low-power magnification of the DLBCL component, seen as multifocal pale eosinophilic areas in a case of mixed DLBCL and MZBL. (B) Follicle colonization by large neoplastic lymphoid cells is seen in the upper corner (right). Diffuse sheeting out is seen in the lower corner (left), separated by a widened lymphatic space.

spectrum described above for the DLBCL component of the mixed DLBCL and MZBL group (Fig. 9). Although less conspicuous than in the MZBL group, at least one morphologic characteristic of lymphomas of MALT-type and 85% of cases (35 ס was seen in 85% of cases (n were associated with lymphocytic thyroiditis. The presence of LEL, follicular colonization, monocytoid B-cell FIG. 6. Low-power magnification of a MZBL with a follic- differentiation, and plasmacytoid differentiation varied ular growth pattern. in quantity and quality from case to case.

Am J Surg Pathol, Vol. 24, No. 5, 2000 630 G. A. DERRINGER ET AL.

FIG. 8. A mixed DLBCL and MZBL showing small plasmacytoid cells with Dutcher bodies (MZBL, A) and large atypical plasmacytoid cells (DLBCL, B).

Follicle Center Lymphoma (FCL). The follicle cen- sive plasmacytoid differentiation (Table 3). None of the effaced the thyroid parenchyma lymphomas that showed a follicular pattern were bcl-2- (1 ס ter lymphoma (n with relatively uniformly sized and shaped nodules com- positive, although 16% of all cases were bcl-2-reactive. prised of small-cleaved lymphocytes with interspersed Aberrant expression of CD43 occurred in 4.4% of cases centroblastic cells. Flattened and atretic thyroid follicular tested (4 of 92). Kappa or lambda light chain restriction epithelium surrounded the neoplastic nodules in a cir- was demonstrated in 19% of cases (Fig. 10). Antibodies cumferential manner. Marginal zone differentiation and for cytokeratin were particularly useful in highlighting LEL were not obvious and were presumed lacking. LEL in areas of DLBCL (Fig. 3). CD10 immunohistochemical analysis was performed on those cases that demonstrated a follicular pattern. Immunoreactivity was Immunohistochemical Analysis uneven with weak staining of small clusters or individual cells in the follicular areas, interpreted as residual reac- The B-cell immunophenotype of the lymphomas was tive germinal center cells, or highlighted the germinal confirmed by immunoreactivity with antibodies to CD20 centers in areas of lymphocytic thyroiditis. The follicle in 105 cases and with CD79a in three cases with exten- center lymphoma was strongly positive for both CD10

FIG. 9. Two different cases exemplify the variable histologic appearance of the diffuse large B-cell lymphomas, monocytoid B-cell-like (A) and plasmacytoid (B).

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 631 and bcl-2. The case that simulated mantle cell lymphoma TABLE 3. Immunohistochemical findings of main and the case that had the appearance of small lympho- histologic categories of PTLs cytic lymphoma were negative for antibodies to CD5, Mixed DLBCL CD23, CD43, and bcl-1. MZBL and MZBL DLBCL without Antibody (N = 30) (N = 36) MZBL

CD45RB 100% (29/29) 97% (33/34) 95% (35/37) Molecular Analysis CD20 93% (28/30) 97% (35/36) 100% (41/41) CD3 0% (0/30) 0% (0/35) 0% (0/36) In molecular analysis, amplifiable DNA was obtained CD45RO 4% (1/26) 9% (3/33) 9% (3/35) in 17 of the 25 lymphomas that demonstrated a follicular CD43 7% (2/28) 6% (2/34) 0% (0/31) CD45RA 40% (10/25) 16% (5/31) 14% (4/29) pattern, at least focally, including seven of eight Bcl-2 18% (5/28) 28% (5/18) 22% (6/27) cases with an extensive follicular pattern that simulated Kappa r 10% (3/29) 9% (3/33) 3% (1/29) FCL. None were positive for either the t(14;18) Lambda r 14% (4/29) 15% (5/33) 3% (1/29) bcl-2/MBR-JH (major break point/MBR) or the t(14;18) MZBL, marginal zone B-cell lymphoma of mucosa-associated bcl-2/MCR-JH (minor cluster region/MCR) assays. The lymphoid tissue type; DLBCL, diffuse large B-cell lymphoma; r, single case classified as follicle center lymphoma was light chain restriction. also negative for both assays. The single case that simulated a mantle cell lymphoma was negative for the t(11;14) assay. slightly shorter duration of symptoms than the other groups, this difference was not statistically significant .(0.241 ס p) Clinical Age, Gender, Signs, and Symptoms Diagnostic Investigations The patients included 79 women and 29 men, with a female to male ratio of 2.7:1 (Table 4). Their ages ranged Radioactive iodine uptake roentgenographic proce- from 28 to 95 years with a mean of 64.3 years at initial dures were performed in the majority of patients, al- presentation. The difference in mean age at presentation though a detailed radiographic narrative was available in for women (65.7 yrs) and men (60.5 yrs) was not statis- only 52. The scans were reported to show “cold” nodules -Curiously, men with in 21 cases, diffuse low-uptake in 19 cases, and no sig .(0.135 ס tically significant (p DLBCL without MZBL presented a decade earlier (54.9 nificant abnormalities in 12 cases. Serum antibody re- yrs) compared with women (67.5 yrs), but this difference sults were available in too few cases to be meaningfully .reported .(0.508 ס was also not statistically significant (p Clinically, all patients presented with a mass in the ס thyroid gland that was noticeably enlarging in 72% (n 78) of cases (Table 4). Rapidly enlarging masses were Stage reported more frequently in DLBCL (29% of DLBCL without MZBL, 28% of mixed DLBCL and MZBL) than The majority of patients presented with stage IE dis- Table 5). An additional 27 patients ;66% ,71 ס in the MZBL (6.6%), yielding a statistically significant ease (n Symptoms resulting from compres- presented with stage IIE (including the one follicular .(0.042 ס result (p sion/infiltration of the neck organs such as dyspnea, dys- lymphoma), with the remaining patients presenting with disease. No (8 ס and stage IVE (n (2 ס phagia, choking and/or coughing, and hemoptysis were stage IIIE (n Pain was also expe- patients with stage IE died with disease, irrespective of .(33 ס reported in 31% of cases (n rienced. Additional symptoms were reported more fre- the histologic grade of the tumor (mean, 82.2 mos), quently in enlarging tumors (37%) or rapidly enlarging whereas 44% of stage IIE, 100% of stage IIIE, and 75% tumors (33%) than in tumors that presented as a mass of stage IVE patients died of their disease. MZBL pa- only (17%) and were significantly more common in tients presented with stage IE (83%) or IIE (17%) dis- As the percentage of large cell ease. All patients who presented with stage IIIE or IVE .(0.048 ס DLBCL (p component increased, there was a greater frequency of had DLBCL. When dissemination was present, either at associated symptoms resulting from compression/ presentation or subsequently, the most common site of infiltration of the neck structures (17% of MZBL, 25% of involvement was lymph nodes. Cervical/perithyroidal mixed DLBCL and MZBL, and 46% of DLBCL without lymph nodes were the most common nodal sites in- MZBL). volved, usually identified at, or shortly after initial sur- Overall, the duration of symptoms ranged from a few gery, followed by mediastinal and abdominal lymph days to 36 months (mean, 4 mos; Table 4). Whereas nodes. Extranodal involvement was found in 8.3% of all -gastrointesti ,(4 ס patients in the DLBCL without MZBL category had a cases and included bone marrow (n

Am J Surg Pathol, Vol. 24, No. 5, 2000 632 G. A. DERRINGER ET AL.

FIG. 10. Light chain restriction for kappa (A) (lambda, B).

The single case classified as FCL was .(45 ס and bladder and liver (42%; n ,(2 ס lung (n ,(3 ס nal tract (n (one each). treated with surgery and chemotherapy only (not included in Table 6). The type of treatment did not correlate with death with Treatment and Clinical Outcome -sur ,(0.098 ס disease, irrespective of surgery alone (p -surgery fol ,(0.639 ס All patients were treated with surgical excision (Table gery followed by radiation (p -or surgery fol ,(0.117 ס including thyroid lobectomy or partial thyroidectomy lowed by chemotherapy (p ,(6 .(0.091 ס lymph node dissection lowed by combination therapy (p (40 ס or without (n (8 ס with (n When controlling for gender, there was no statistical ס or without (n (11 ס and total thyroidectomy with (n -duration of symp ,(0.508 ס lymph node dissection. In addition to surgery, adju- association among age (p (49 ס or maximum size of tumor (p ,(0.822 ס vant therapy was used in 75% of patients (81 of 108), toms (p commonly a combination of radiation and chemotherapy 0.694) and evidence of disease at last follow up. Al-

TABLE 4. Clinical characteristics of main histologic categories of PTLs

Mixed DLBCL DLBCL All MZBL and MZBL without MZBL Clinical finding PTLs* (N = 30) (N = 36) (N=41)

Total number 108 30 36 41 Gender Women 79 20 25 34 Men 29 10 11 7 Age Range 28–95 yrs 34–84 yrs 30–95 yrs 28–93 yrs Mean 64.3 yrs 63.1 yrs 64.6 yrs 65.3 yrs Women (mean) 65.7 yrs 63.2 yrs 65.3 yrs 67.5 yrs Men (mean) 60.5 yrs 63.0 yrs 62.9 yrs 54.9 yrs Symptoms Mass 30 (28%) 11 (37%) 9 (25%) 9 (22%) Enlarging mass 54 (50%) 17 (57%) 17 (47%) 20 (49%) Rapidly enlarging mass 24 (22%) 2 (7%) 10 (28%) 12 (29%) Additional symptoms Dysphagia 18 (17%) 4 (13%) 3 (8%) 11 (27%) Cough 13 (12%) 2 (7%) 5 (14%) 6 (15%) Dyspnea 11 (10%) 1 (3%) 3 (8%) 7 (17%) Hoarseness 2 (1.9%) 0 (0%) 0 (0%) 2 (4.9%) Hemoptysis 1 (<1%) 0 (0%) 0 (0%) 1 (2.4%) Duration of symptoms Range 0–36 mos 0–24 mos 0–36 mos 0–8 mos Mean 4.0 mos 4.7 mos 5.0 mos 3.1 mos

PTLs, primary thyroid gland lymphomas; MZBL, marginal zone B-cell lymphoma of mucosa- associated lymphoid tissue type; DLBCL, diffuse large B-cell lymphoma. * Includes the single follicle center lymphoma.

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 633

TABLE 5. Clinical stage and patient outcome (at last follow up) of main histologic categories of PTLs

Mixed DLBCL DLBCL without MZBL and MZBL MZBL N=30(avgmos) N=36(avgmos) N=41(avgmos)

Stage DWD NED DWD NED DWD NED

IE — 25 (92.3) — 22 (74.7) — 24 (80.8) IIE — 5 (84.5) 4 (12.1) 4 (104.0) 8 (21.6) 5 (67.7) IIIE ——1 (29.4) — 1 (88.5) — IVE ——3 (19.3) 2 (72.5) 3 (13.6) —

PTLs, primary thyroid gland lymphomas; MZBL, marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue type; DLBCL, diffuse large B-cell lymphoma; avg mos, average number of months of follow up; DWD, dead with disease; NED, no evidence of disease, whether alive or dead of unrelated causes. though patients who presented with rapidly growing tu- cant difference in survival between those patients who died with disease and those patients who were alive or ,(0.002 ס mors were more likely to die with disease (p -The disease .(0.0001 ס they did not have an increased likelihood of extrathyroi- had died of unrelated causes (p ,specific 5-year survival rates by category were: MZBL .(0.065 ס dal extension (p Overall, 70 of 108 (65%) patients were alive at last 100%; mixed DLBCL and MZBL, 78%; and DLBCL follow up, which translates into a raw survival of 60% at without MZBL, 71%, whereas the overall 5-year survival 5 years (14 additional patients were alive but were not rates by category were slightly shorter as a result of yet followed for 5 yrs). However, 88 of 108 (82%) pa- intercurrent, but unrelated, deaths (MZBL, 77%; mixed tients were alive without evidence of disease, or had died DLBCL and MZBL, 47%; and DLBCL without MZBL, of other causes, with a mean follow up of 82.8 months. 54%). The patient with the tumor classified as FCL was Thus, at 5 years, the disease-specific survival was 79% alive without disease, almost 4 years after diagnosis. (74 of 94 patients who had been followed for a total of 5 All patients who died with disease presented with years were either alive or had died of other causes). stage IIE or higher disease and had DLBCL. Stage of Twenty patients died with disease (mean, 21.9 mos), disease was strongly associated with death with disease. with most deaths (90%) occurring within 3 years of di- The strong statistical significance is the result of the ס agnosis. All had DLBCL. Of particular interest in our difference between stage IE and stage IIE groups (p series: if the patient was alive at last follow up, they were 0.0001; Table 7). Statistical difference between stage IIE disease-free. Eighteen patients died without evidence of and combined stage IIIE and stage IVE groups was not .(0.133 ס disease of unrelated causes, including 7 patients with significant (p MZBL, 7 with mixed DLBCL and MZBL, and 4 with The presence of a DLBCL was strongly associated DLBCL without MZBL. There was a statistically signifi- with death with disease and DLBCL had a worse clinical

TABLE 6. Overall patient outcome at last follow-up by treatment regimen for main histologic categories of PTLs

Mixed DLBCL DLBCL without MZBL and MZBL MZBL N=30(avgmos) N=36(avgmos) N = 41 (avg mos)

DWD NED DWD NED DWD NED

Survival Overall 0 30 (91.0) 8 (17.0) 28 (80.4) 12 (25.2) 29 (79.3) Range in months N/A 12–139 0.5–35 2–162 0.5–89 7–149 Type of treatment Surgery only 0 16 2 6 1 2 (84.1) (15.0) (56.4) (0.1) (23.3) Surgery and radiation therapy 0 5 2 8 1 4 (105.6) (14.6) (93.5) (4.8) (71.4) Surgery and chemotherapy 0 3 2 3 3 4 (99.0) (13.1) (111.9) (44.5) (111.9) Surgery and combined therapy 0 6 2 11 7 19 (93.5) (25.2) (78.3) (23.3) (77.4)

Am J Surg Pathol, Vol. 24, No. 5, 2000 634 G. A. DERRINGER ET AL.

TABLE 7. Statistical significance of clinical and by the tumor. With radioisotopic 131I scanning, a cold histologic parameters for patient outcome using dead nodule or areas of decreased uptake are common. with disease as the end point of PTLs PTLs demonstrate considerable diversity in architec- Variable p value ture and cytology, not only from case to case, but within the same case. Although isolated T-cell PTLs have been Stage greater than IE 0.0001 1,54,59,69,82 Diffuse large cell lymphomas 0.002 reported, all cases in this study were B-cell Rapid clinical growth 0.002 lymphomas. Barring the single FCL, they fell into three Abundant apoptosis 0.003 Presence of vascular invasion 0.005 groups: MZBL, mixed DLBCL and MZBL, and DLBCL High mitotic rate 0.022 without MZBL. Many of our cases (71%) contained a Presence of perithyroidal extension 0.025 DLBCL element, paralleling the findings in the litera- Symptoms of compression/infiltration 9,11,13,14,23–25,34,36,41,42,51,52,54,59,64,69–71,75–78 of neck organs by tumor mass 0.040 ture. Con- Modalities of treatment 0.05* versely, almost half (47%) of our tumors that contained Degree of fibrosis 0.123* a DLBCL contained a MZBL, to which the term mixed Presence of necrosis 0.138* Duration of symptoms 0.265* DLBCL and MZBL was applied (in keeping with termi- Maximum tumor size 0.346* nology of the REAL and proposed WHO classifications). Age 0.61* The small and large cell components varied in proportion PTLs, primary thyroid gland lymphomas. from case to case, requiring evaluation of several slides * Not statistically significant. to recognize the mixture. Random microscopic foci of DLBCL or small areas of MZBL could be easily missed on cursory examination or with limited sampling. Mixed Table 7). No patients with MZBL ;0.002 ס outcome (p DLBCL and MZBL, previously reported in other extra- died with disease (mean, 91.0 mos). All 20 patients who nodal sites17,22,30,46,60,61,66 as well as in the thyroid died with disease had DLBCL (eight patients with mixed gland,34,45,59,69 seem to represent approximately one DLBCL and MZBL, mean, 17.0 mos; 12 patients with third of all thyroid lymphomas, similar to our findings DLBCL without MZBL, mean, 25.2 mos) with no dif- (33% of all cases). Could the mixed DLBCL and MZBL ference between mixed DLBCL and MZBL and DLBCL cases represent de novo mixed small and large cell pre- .(0.604 ס without MZBL groups (p sentations? Absolutely, but the occasional focal nature of Perithyroidal soft tissue invasion was significantly as- the components, areas of transition between the two his- as 18 of 20 ,(0.025 ס sociated with death with disease (p tologies, and the clinical presentation of a rapidly enlarg- patients (90%) who died of disease had lymphomas that ing mass lend support to the transformation theory (large demonstrated perithyroidal extension. Likewise, the cell transformation). Furthermore, many patients with presence of vascular invasion was significantly associ- PTLs have a history of sustained thyroid enlargement or ,Furthermore .(0.005 ס ated with death with disease (p goiter before the presentation of a mass or rapid enlarge- vascular invasion was only identified in DLBCL, and ment of the gland, suggesting that there may have been a those who died with disease had stage IIE or higher pre-existing MZBL before the development of DLBCL disease. Patients with lymphomas that demonstrated (large cell transformation).3,5,6,9,11,19,23,42,71,78 In fact, a and high mitotic rates (0.003 ס abundant apoptosis (p study of a few patients with a history of persistent goiter histologic features more often found in the ,(0.022 ס p) retrospectively had MZBL that went undetected.34 More- were significantly more likely to ,(0.0001 ס DLBCL (p over, immunohistochemical and molecular studies char- die with disease (Table 7). Although tumor necrosis was acterizing mixed DLBCL and MZBL in the thyroid and significantly more likely to be present in the DLBCL stomach lend further support to the transformation this histologic finding was not associated ,(0.0001 ס p) theory.17,22,30,60,61 Similarly, grade of .(0.138 ס with death with disease (p Although characteristics of lymphomas of MALT- ס fibrosis was not associated with death with disease (p type were more obvious in MZBL, they were also found 0.123). There was no significant difference in the num- in the areas of DLBCL. Barring lymphocytic thyroiditis, ber of sections examined per case between the tumor LEL were the most common characteristic identified in .(0.664 ס categories (p the DLBCL, accentuated with CD20 and cytokeratin stains. Although there were notable exceptions, the neo- DISCUSSION plastic large lymphocytes were more destructive and in- vasive. Consequently, the LEL were typically fewer, PTLs typically occur in middle- to older-aged indi- smaller, and more inconspicuous than in the MZBL areas viduals in the setting of lymphocytic thyroiditis with a and were most often of the usual type. In our series, the predilection for females. Patients present with a mass in “MALT ball” type of LEL, seen most often in MZBL, the thyroid, usually noticeably enlarging, and commonly were unique and seemed specific for malignancy, a fea- have symptoms related to compression of neck structures ture that has been suggested by others.3,5,9,19,31,56 We

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 635 attribute the differences in quality and quantity of LEL tern. None of these cases were immunoreactive for CD10 between the MZBL and DLBCL components to the tu- (on frozen instead of paraffin-embedded, formalin-fixed mor’s degree of differentiation.38 In a large cell lym- tissue) or showed evidence of a t(14;18) using polymer- phoma, the presence of a MZBL and/or LEL has been ase chain reaction on DNA from fresh and/or paraffin- used by some authors to classify a lymphoma as a lym- embedded tissue.37 Similar findings, using immunohis- phoma of MALT-type.30,59 While LEL of the usual type tochemical and molecular techniques, have been reported are not specific and can be seen in benign lymphoid in gastrointestinal tract MZBL which demonstrated a fol- proliferations involving the thyroid such as Hashimoto’s licular pattern.39,58 From a clinical perspective, the tu- (lymphocytic) thyroiditis,31,47,50 LEL formed by either mors in our study were not characteristic for FCL, almost small or large B-lymphocytes which diffusely efface the always presenting with low-stage disease without bone thyroid parenchyma support a lymphoma of MALT-type marrow involvement. Only one patient presented with in the thyroid. In other words, a DLBCL in the thyroid high-stage disease (IVE), with the lymphoma identified gland that demonstrates LEL, whether or not there is an in the bladder and gastrointestinal tract, organs that can accompanying MZBL, supports that it is a lymphoma of be considered part of the “MALT axis.” Furthermore, the MALT-type. However, because we think the ability of a outcomes were not characteristic of FCL. Of the patients lymphoma of MALT-type to form LEL in the thyroid is with a follicular pattern, only two patients died with dis- affected by its degree of differentiation, a lack of promi- ease less than a year after presentation (mixed DLBCL nent LEL should not preclude the same designation. and MZBL, initially stage IE disease). The remaining (2 ס or had died (n (21 ס Lymphocytic thyroiditis is almost certainly a requisite patients were either alive (n for the development of lymphoma in the thyroid. The of other causes with no evidence of disease at last follow few cases (<7%) that did not have documented lympho- up. Our only case of FCL did not yield positive t(14;18) cytic thyroiditis may have been either inadequately assays. However, the negative t(14;18) assay could be sampled or the lymphoma may have completely obliter- explained by a false-negative result. On the other hand, ated any residual thyroid tissue. Furthermore, the chro- there is room for doubt about our classification of this nicity of the disease process was supported by the pres- case, because the patient presented with stage IIE dis- ence of atrophy of the thyroid parenchyma and the pres- ease, had lymphocytic thyroiditis, and is alive without ence of fibrosis, often of a moderate to severe degree, in evidence of disease at last follow up, 46.5 months after all cases. presentation. In this study, FCL, as a PTL, is uncommon when It is our supposition that most PTL are lymphomas of defined by the clinicopathologic features typical of those MALT-type that have a highly variable cytomorphologic arising in nodal sites. Of 108 cases, we made this diag- appearance and range from MZBL to “DLBCL of nosis in only one case, based on histologic and im- MALT-type.” However, they can simulate “non-MALT- munohistochemical findings. The FCL in our series dem- type” lymphomas.6 The architectural heterogeneity of onstrated a prominent follicular architecture, lacked de- lymphomas in the thyroid may depend on the mutational finitive features of a MZBL (although there was lym- events leading to the development of the neoplastic phocytic thyroiditis), and was immunoreactive for CD10. clone. It is well known that centrocyte-like lymphocytes Before the understanding of the MALT concept, a sig- are common components of the typically cytologically nificant proportion of PTLs was thought to be of folli- heterogeneous MZBL. Is it possible that the neoplastic cle center origin.3–5,8,9,11,19,23,28,48,50,52,54,56,70 It is well clone of some lymphomas that develop out of MALT known that extensive colonization of the germinal cen- favors follicle center or mantle zone-like differentiation ters of reactive follicles by neoplastic marginal zone over monocytoid B-cell or plasma cell lines? This theory cells in MZBL can resemble follicle center lym- is supported by the recent description of a nodular vari- phoma.34,36–38,46 The follicular pattern seen in 25 of our ant of marginal zone B-cell lymphoma that can simulate cases could be attributed to this phenomenon, a finding follicle center or even a mantle cell lymphoma.15 Al- -though further study is necessary, we think, in the thy ס similar to that of other authors.3 In many of these (n 17), features of MZBL were easily identified, raising roid, FCLs are uncommon and MCLs are exceptional. little question as to their subclassification. However, we Because appropriate treatment and prognostic implica- were impressed how eight cases closely resembled FCL. tions depend on accurate classification, we recommend On closer observation, despite a prominent nodular ar- that the diagnosis of a primary FCL or MCL in the thy- chitecture, marginal zone differentiation and LEL could roid be made only when the tumor arises primarily in the be distinguished. Additionally, all were negative for thyroid and there is supportive immunohistochemical bcl-2 and CD10 by immunohistochemical analysis and and/or molecular evidence. were negative for t(14;18) molecular assays. A study Plasma cells and plasmacytic differentiation specifically addressing this issue analyzed nine MZBL of were common in our PTL. It is not difficult to sug- the thyroid that demonstrated a prominent follicular pat- gest as others have3,33,36 that cases previously report-

Am J Surg Pathol, Vol. 24, No. 5, 2000 636 G. A. DERRINGER ET AL. ed as extramedullary plasmacytomas of the thy- because of differences in population size, study design, roid4,7,9,12,18,19,23,44,54,57,68,80 represent MZBL with ex- lymphoma classification used, treatment modalities used, treme plasma cell differentiation. Although a few of our and statistical analyses used. As expected,14,19,76,80 a MZBL cases could have been called extramedullary number of clinical and histologic findings were pre- plasmacytomas (EMP) by some pathologists, studies dictive of patient outcome (Table 7), although the rela- support that EMP are different entities from solitary plas- tionship of stage of disease and histologic grade are macytomas of the bone (SPB) and multiple myeloma certainly interrelated. Patients who presented with en- (MM) (neoplasms arising from marrow cells) and that larging tumors or symptoms of compression of neck EMP are probably MZBL with extensive plasma cell organs were more likely to die with disease, as ex- 20,33,53,79 differentiation with origin from MALT. This pected.13,41,48,69,71,76 However, age at presentation, gen- theory is based on the numerous clinicopathologic simi- der, and size of the tumor were not prognostic markers, larities that exist between EMP and MZBL. In our study, contrary to other studies.5,41,43,48,71,75,76 we found that the quantity of the plasma cell component The majority (91%) of our patients presented with varied greatly, as did other histologic characteristics of stage IE or IIE disease, analogous to other re- MZBL, from one case to the next, thus indicating a dis- ports.5,9,13,14,23,41,45,48,51,54,59,62,69,71,74,75 As noted in ease with a wide histologic spectrum. Although not all other studies,3,14,19,23,42,45,48,54,56,59,63,69,71 cervical characteristics of MZBL were seen in our plasma cell- lymph nodes were the most frequently involved site, fol- predominant MZBL cases, many features could be iden- lowed by a variety of other nodal and extranodal sites, tified. These findings are similar to a recent study ad- including mediastinal lymph nodes, gastrointestinal tract, dressing this issue in which features of MZBL could be bone marrow, lung, bladder, liver, and spleen. Stage of identified in five cases from various non-thyroidal extra- disease is an important prognostic indicator, with stage IE medullary sites that were diagnosed as EMP originally yielding an excellent clinical outcome. Stage is an im- but thought to represent MZBL with extensive plasma portant prognostic indicator,13,14,19,41,48,52,54,59,71 although cell differentiation.33 Although EMP in the thyroid could others have not found this to be the case.5,62,74–76,80 We represent different entities from plasma cell predominant realize that differences in staging modalities and vari- MZBL, mounting evidence appears to support otherwise. ability in treatment regimens can affect findings. How- Lastly, plasma cell dyscrasias such as MM could involve ever, none of the patients with stage IE, irrespective of the thyroid secondarily, but clear-cut evidence fulfilling histologic grade, died with disease. Therefore, because of established diagnostic criteria to support this unusual event (that is, atypical plasma cell infiltrate of the mar- prognostic implications, accurate staging is important. row, lytic bone lesions, monoclonal gammopathy) would It is difficult to separate clinical stage and histologic be required to support this interpretation. findings as prognostic indicators, because these param- The plasma cells in a MZBL of the thyroid may be eters are intimately interrelated. In our study, only pa- reactive and part of the lymphocytic thyroiditis. This tients with DLBCL (mixed DLBCL and MZBL and may account for our inability to demonstrate mono- DLBCL without MZBL) presented with advanced stage clonality with immunohistochemistry in a number of (stage IIIE or IVE) and/or had a poor outcome. All our cases. However, CD79a immunoreactivity was MZBL presented with lower stage disease (most stage helpful in confirming the B-cell immunophenotype of IE) and none died with disease. Although evidence to cases that were negative for L26 because of plasmacy- support histology as a prognostic indicator varies in the toid differentiation. While there may be true T-cell lym- literature, in our study, the lymphomas, when divided phomas of MALT origin of the thyroid,38 because into two main subgroups MZBL and DLBCL, demon- UCHL-1 can be expressed by some B-cell lymphomas, strated a strong statistical association with death with Therefore, the distinction between .(0.002 ס as observed in our study, it should not be used alone. The disease (p use of more specific T-cell markers such as CD3 and MZBL and DLBCL in the thyroid is important and clini- CD5 is recommended. cally significant. Distinction between mixed DLBCL and Reed-Sternberg-like cells are known to occur in a MZBL and DLBCL without MZBL categories appears wide variety of lymphoma subtypes and in reactive lym- not to be important. However, analysis of a larger series phoid proliferations, so it is not surprising that a few of of cases may be necessary to confirm this conclusion. our cases, like those in the literature,14,52,80 contained Because the majority (71%) of our patients with these cells. However, we could not verify the claims of DLBCL had a favorable outcome, we studied additional others that report primary thyroid Hodgkin’s lymphoma histologic factors to see if they had an influence on clini- in our series.19,26,42,49,76 cal behavior. In our series, presence of vascular invasion, It is difficult to compare our 5-year disease-specific survival abundant apoptosis, high mitotic rate, and perithyroidal of 79% to the wide range of 35% to 79% reported in the soft tissue invasion were significant histologic prognos- literature3,5,11,13,14,19,24,41,42,45,48,51,52,54,59,62,64,69–72,74–77,80 tic indicators. Vascular invasion5,19,56 and perithyroidal

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 637 invasion11,14,19,42,56,76,78,80 have been demonstrated to be clinical stage, with patients having MZBL and stage indicators of poor clinical outcome in the past. There is IE disease yielding the best prognosis and those with no study, to our knowledge, that has demonstrated ap- DLBCL or greater than stage IE disease having the great- optosis as a significant prognostic factor. Although all of est potential for a poor outcome. Further study is needed these histologic features were significantly more charac- to determine the occurrence and frequency of PTLs of teristic of DLBCL (high mitotic activity, frank necrosis, “non-MALT-type.” ᮀ with the exception ,(0.0001 ס and abundant apoptosis; p of vascular invasion, these histologic features could be Acknowledgments identified without adverse effects on clinical outcome in The authors thank Robin-Anne Ferris and Luther Duckett for the MZBL group. However, there may be use in seeking their expert photography, Ian W. McLean and Allen Burke for these histologic features, particularly in DLBCL, be- their statistical analysis, and Pamela Thompson for her research cause their presence may portend a greater risk of recur- assistance. rence and/or dying of disease. REFERENCES As expected, chemotherapy was used in a greater per- 1. Abdul-Rahman ZH, Gogas HJ, Tooze JA, et al. T-cell lymphoma centage of the DLBCL, but it is difficult to accurately in Hashimoto’s thyroiditis. Histopathology 1996;29:455–9. determine the impact of a single therapeutic modality on 2. Aguilera NS, Bijwaard KE, Duncan B, et al. Differential expres- patient survival when a wide variety of different treat- sion of cyclin D1 in mantle cell lymphoma and other non- Hodgkin’s lymphomas. Am J Pathol 1998;153:1969–76. ment modalities were applied in an uncontrolled manner. 3. Anscombe AM, Wright DH. Primary malignant lymphoma of the Therefore, we cannot extrapolate our findings to clinical thyroid—a tumour of mucosa-associated lymphoid tissue: review application. However, authors have documented a sig- of seventy-six cases. Histopathology 1985;9:81–97. 4. Aozasa K, Inoue A, Katagiri S, et al. Plasmacytoma and follicular nificant relapse rate of PTLs at distant sites, attributed to lymphoma in a case of Hashimoto’s thyroiditis. Histopathology treatment failure.13,21,36,41,63,75–77 Whereas a number of 1986;10:735–40. studies have reported improved survival with tumor de- 5. Aozasa K, Inoue A, Tajima K, et al. Malignant lymphomas of the 13,41,67,75,76 thyroid gland. Analysis of 79 patients with emphasis on histologic bulking and local control of the tumor, a more prognostic factors. Cancer 1986;58:100–4. favorable patient outcome seems to follow the use of 6. Aozasa K, Inoue A, Yoshimura H, et al. Intermediate lymphocytic combined adjuvant therapy,51,63,71,75 even without tumor lymphoma of the thyroid. An immunologic and immunohistologic study. Cancer 1986;57:1762–7. debulking. Our excellent overall survival rates may be 7. Aozasa K, Inoue A, Yoshimura H, et al. Plasmacytoma of the related to a combination of good local control of the thyroid gland. Cancer 1986;58:105–10. tumor through the use of surgery and radiation therapy 8. Aozasa K, Nara H, Ikeda H, et al. The influence of histologic type on survival in early extranodal non-Hodgkin’s lymphoma in head and control of disseminated disease, undetected at pre- and neck. Oncology 1984;41:164–9. sentation, with chemotherapy (57% of patients), despite 9. Aozasa K, Tajima K, Tominaga N, et al. Immunologic and immu- a lower stage of disease at presentation. Chemotherapy nohistochemical studies on chronic lymphocytic thyroiditis with or without thyroid lymphoma. Oncology 1991;48:65–71. may have treated undetected occult disseminated disease 10. Aozasa K, Tsujimoto M, Sakurai M, et al. Non-Hodgkin’s lym- in our patients and protected against recurrence, particu- phomas in Osaka, Japan. Eur J Cancer Clin Oncol 1985;21:487– larly for those patients with DLBCL. Further, recent im- 92. 11. Aozasa K, Ueda T, Katagiri S, et al. Immunologic and immuno- provements in radiotherapeutic techniques and better histologic analysis of 27 cases with thyroid lymphomas. Cancer chemotherapy regimens may account for the better sur- 1987;60:969–73. vival in our patients compared with patients in earlier 12. Beguin Y, Boniver J, Bury J, et al. Plasmacytoma of the thyroid: a case report, a study with use of the immunoperoxidase technique, studies. and a review of the literature. Surgery 1987;101:496–500. In conclusion, we studied 108 PTL using recent con- 13. Blair TJ, Evans RG, Buskirk SJ, Banks PM, Earle JD. Radiothera- cepts. PTLs typically occur in middle- to older-aged per- peutic management of primary thyroid lymphoma. Int J Radiat Oncol Biol Phys 1985;11:365–70. sons and have a predilection for females. Virtually all 14. Burke JS, Butler JJ, Fuller LM. Malignant lymphomas of the thy- PTLs are lymphomas of MALT-type, arising as neoplas- roid: a clinical pathologic study of 35 patients including ultrastruc- tic transformations from MALT in the form of lympho- tural observations. Cancer 1977;39:1587–602. 15. Campo E, Miquel R, Krenacs L, et al. Primary nodal marginal zone cytic thyroiditis. In keeping with the current terminol- lymphomas of splenic and MALT type. Am J Surg Pathol 1999; ogy, most PTLs can be classified as MZBL or DLBCL 23:59–68. (DLBCL of MALT-type). Mixed DLBCL and MZBL are 16. Carbone PP, Kaplan HS, Musshoff K, Smithers DW, Tubiana M. Report of the Committee on Hodgkin’s Disease Staging Classifi- common. Because the histology of PTL varies tremen- cation. Cancer Res 1971;31:1860–1. dously and can sometimes simulate other “non-MALT- 17. Chan JK, Ng CS, Isaacson PG. Relationship between high-grade type lymphomas,” specifically FCL, familiarity with the lymphoma and low-grade B-cell mucosa-associated lymphoid tissue lymphoma (MALToma) of the stomach. Am J Pathol 1990; various patterns of PTL will prevent misclassification 136:1153–64. and clinical confusion. The overall clinical outcome for 18. Chesyln-Curtis S, Akosa AB. Primary plasmacytoma of the thy- PTL is favorable with appropriate management, but re- roid. Postgrad Med J 1990;66:477–8. 19. Compagno J, Oertel JE. Malignant lymphoma and other lympho- currence and death from disease do occur. The prognosis proliferative disorders of the thyroid gland. A clinicopathologic of PTL is dependent on both histologic findings and study of 245 cases. Am J Clin Pathol 1980;74:1–11.

Am J Surg Pathol, Vol. 24, No. 5, 2000 638 G. A. DERRINGER ET AL.

20. Corwin J, Lindberg RD. Solitary plasmacytoma of bone vs. extra- of the thyroid: report of a case of solitary plasmacytoma and a medullary plasmacytoma and their relationship to multiple my- review of the literature. Thyroid 1994;4:65–71. eloma. Cancer 1979;43:1007–13. 45. Laing RW, Hoskin P, Hudson BV, et al. The significance of 21. Doria R, Jekel JF, Cooper DL. Thyroid lymphoma. The case for MALT histology in thyroid lymphoma: a review of patients from combined modality therapy. Cancer 1994;73:200–6. the BNLI and Royal Marsden Hospital. Clin Oncol (R Coll Radiol) 22. Du M, Peng H, Singh N, Isaacson PG, Pan L. The accumulation of 1994;6:300–4. p53 abnormalities is associated with progression of mucosa- 46. Li G, Hansmann ML, Zwingers T, Lennert K. Primary lymphomas associated lymphoid tissue lymphoma. Blood 1995;86:4587–93. of the lung: morphological, immunohistochemical and clinical fea- 23. Faure P, Chittal S, Woodman-Memeteau F, et al. Diagnostic features. Histopathology 1990;16:519–31. tures of primary malignant lymphomas of the thyroid with mono- 47. Li VV. The pathology of autoimmune thyroid disease: a review. clonal antibodies. Cancer 1988;61:1852–61. Thyroid 1994;4:333–9. 24. Freeman C, Berg JW, Cutler SJ. Occurrence and prognosis of 48. Logue JP, Hale RJ, Stewart AL, Duthie MB, Banerjee SS. Primary extranodal lymphomas. Cancer 1972;29:252–60. malignant lymphoma of the thyroid: a clinicopathological analysis. 25. Hamburger JI, Miller JM, Kini SR. Lymphoma of the thyroid. Ann Int J Radiat Oncol Biol Phys 1992;22:929–33. Intern Med 1983;99:685–93. 49. Luboshitzky R, Dharan M, Nachtigal D, Rosen G, Honigman J. 26. Hardoff R, Bar-Shalom R, Dharan M, Luboshitsky R. Hodgkin’s Syncytial variant of nodular sclerosing Hodgkin’s disease present- disease presenting as a solitary thyroid nodule. Clin Nucl Med ing as a thyroid nodule. A case report. Acta Cytol 1995;39:543–6. 1995;20:37–41. 50. Matias-Guiu X, Esquius J. Lymphoepithelial lesion in the thyroid. 27. Harris NL, Jaffe ES, Stein H, et al. A revised European-American A non-specific histological finding. Pathol Res Pract 1991;187: classification of lymphoid neoplasms: a proposal from the Inter- 296–300. national Lymphoma Study Group. Blood 1994;84:1361–92. 51. Matsuzuka F, Miyauchi A, Katayama S, et al. Clinical aspects of 28. Heimann R, Vannineuse A, De Sloover C, Dor P. Malignant lym- primary thyroid lymphoma: diagnosis and treatment based on our phomas and undifferentiated small cell carcinoma of the thyroid: a experience of 119 cases. Thyroid 1993;3:93–9. clinicopathological review in the light of the Kiel classification for 52. Maurer R, Taylor CR, Terry R, Lukes RJ. Non-Hodgkin lympho- malignant lymphomas. Histopathology 1978;2:201–13. mas of the thyroid. A clinico-pathological review of 29 cases ap- 29. Holm LE, Blomgren H, Lowhagen T. Cancer risks in patients with plying the Lukes-Collins classification and an immunoperoxidase chronic lymphocytic thyroiditis. NEnglJMed1985;312:601–4. method. Virchows Arch 1979;383:293–317. 30. Hsi ED, Eisbruch A, Greenson JK, et al. Classification of primary 53. Meis JM, Butler JJ, Osborne BM, Ordonez NG. Solitary plasma- gastric lymphomas according to histologic features. Am J Surg cytomas of bone and extramedullary plasmacytomas. A clinico- Pathol 1998;22:17–27. pathologic and immunohistochemical study. Cancer 1987;59: 31. Hsi ED, Singleton TP, Svoboda SM, Schnitzer B, Ross CW. Char- 1475–85. acterization of the lymphoid infiltrate in Hashimoto thyroiditis by 54. Mizukami Y, Michigishi T, Nonomura A, et al. Primary lymphoma immunohistochemistry and polymerase chain reaction for immu- of the thyroid: a clinical, histological and immunohistochemical noglobulin heavy chain gene rearrangement. Am J Clin Pathol study of 20 cases. Histopathology 1990;17:201–9. 1998;110:327–33. 55. Ngan BY, Nourse J, Cleary ML. Detection of chromosomal trans- 32. Hsu SM, Raine L, Fanger H. Use of avidin-biotin-peroxidase com- location t(14; 18) within the minor cluster region of bcl-2 by poly- plex (ABC) in immunoperoxidase techniques: a comparison be- merase chain reaction and direct genomic sequencing of the enzy- tween ABC and unlabeled antibody (PAP) procedures. J Histo- matically amplified DNA in follicular lymphomas. Blood 1989;73: chem Cytochem 1981;29:577–80. 1759–62. 33. Hussong JW, Perkins SL, Schnitzer B, Hargreaves H, Frizzera G. 56. Oertel JE, Heffess CS. Lymphoma of the thyroid and related dis- Extramedullary plasmacytoma. A form of marginal zone cell lym- orders. Semin Oncol 1987;14:333–42. phoma? Am J Clin Pathol 1999;111:111–6. 57. Ohshima M, Momiyama T, Souda S, et al. Primary plasmacytoma 34. Hyjek E, Isaacson PG. Primary B cell lymphoma of the thyroid and of the thyroid: a case report and comparative literature study be- its relationship to Hashimoto’s thyroiditis. Hum Pathol 1988;19: tween Western nations and Japan. Pathol Int 1994;44:645–51. 1315–26. 58. Pan L, Diss TC, Cunningham D, Isaacson PG. The bcl-2 gene in 35. Isaacson P, Wright DH. Malignant lymphoma of mucosa- primary B cell lymphoma of mucosa-associated lymphoid tissue associated lymphoid tissue. A distinctive type of B-cell lymphoma. (MALT). Am J Pathol 1989;135:7–11. Cancer 1983;52:1410–6. 59. Pedersen RK, Pedersen NT. Primary non-Hodgkin’s lymphoma of 36. Isaacson PG. Lymphoma of the thyroid gland. Curr Top Pathol the thyroid gland: a population based study. Histopathology 1996; 1997;91:1–14. 28:25–32. 37. Isaacson PG, Androulakis-Papachristou A, Diss TC, Pan L, Wright 60. Peng H, Chen G, Du M, et al. Replication error phenotype and p53 DH. Follicular colonization in thyroid lymphoma. Am J Pathol gene mutation in lymphomas of mucosa-associated lymphoid tis- 1992;141:43–52. sue. Am J Pathol 1996;148:643–8. 38. Isaacson PG, Spencer J. Malignant lymphoma of mucosa- 61. Peng H, Du M, Diss TC, Isaacson PG, Pan L. Genetic evidence for associated lymphoid tissue. Histopathology 1987;11:445–62. a clonal link between low and high-grade components in gastric 39. Isaacson PG, Wotherspoon AC, Diss T, Pan LX. Follicular colo- MALT B-cell lymphoma. Histopathology 1997;30:425–9. nization in B-cell lymphoma of mucosa-associated lymphoid tis- 62. Pledge S, Bessell EM, Leach IH, et al. Non-Hodgkin’s lymphoma sue. Am J Surg Pathol 1991;15:819–28. of the thyroid: a retrospective review of all patients diagnosed in 40. Jaffe ES, Harris NL, Diebold J, Muller-Hermelink HK. World Nottinghamshire from 1973 to 1992. Clin Oncol (R Coll Radiol) Health Organization classification of neoplastic diseases of the 1996;8:371–5. hematopoietic and lymphoid tissues. A progress report. AmJClin 63. Pyke CM, Grant CS, Habermann TM, et al. Non-Hodgkin’s lym- Pathol 1999;111:S8–12. phoma of the thyroid: is more than biopsy necessary? World J Surg 41. Junor EJ, Paul J, Reed NS. Primary non-Hodgkin’s lymphoma of 1992;16:604–9. the thyroid. Eur J Surg Oncol 1992;18:313–21. 64. Rasbach DA, Mondschein MS, Harris NL, Kaufman DS, Wang 42. Kapadia SB, Dekker A, Cheng VS, Desai U, Watson CG. Malig- CA. Malignant lymphoma of the thyroid gland: a clinical and nant lymphoma of the thyroid gland: a clinicopathologic study. pathologic study of twenty cases. Surgery 1985;98:1166–70. Head Neck Surg 1982;4:270–80. 65. Reid AH, Cunningham RE, Frizzera G, O’Leary TJ. bcl-2 rear- 43. Kato I, Tajima K, Suchi T, et al. Chronic thyroiditis as a risk factor rangement in Hodgkin’s disease. Results of polymerase chain re- of B-cell lymphoma in the thyroid gland. Jpn J Cancer Res 1985; action, flow cytometry, and sequencing on formalin-fixed, paraf- 76:1085–90. fin-embedded tissue. Am J Pathol 1993;142:395–402. 44. Kovacs CS, Mant MJ, Nguyen GK, Ginsberg J. Plasma cell lesions 66. Rivas C, Echezarreta G, Garcia R, et al. A multiparametric study

Am J Surg Pathol, Vol. 24, No. 5, 2000 THYROID LYMPHOMAS 639

of malignant lymphoma of mucosa associated lymphoid tissue Hodgkin’s lymphoma of the thyroid gland: prognostic factors and (MALT). Leuk Lymphoma 1992;8:87–96. treatment outcome. The Princess Margaret Hospital Lymphoma 67. Rosen IB, Sutcliffe SB, Gospodarowicz MK, Chua T, Simpson Group. Int J Radiat Oncol Biol Phys 1993;27:599–604. WJ. The role of surgery in the management of thyroid lymphoma. 76. Tupchong L, Hughes F, Harmer CL. Primary lymphoma of the Surgery 1988;104:1095–9. thyroid: clinical features, prognostic factors, and results of treat- 68. Rubin J, Johnson JT, Killeen R, Barnes L. Extramedullary plas- ment. Int J Radiat Oncol Biol Phys 1986;12:1813–21. macytoma of the thyroid associated with a serum monoclonal gam- 77. Vigliotti A, Kong JS, Fuller LM, Velasquez WS. Thyroid lympho- mopathy. Arch Otolaryngol Head Neck Surg 1990;116:855–9. mas stages IE and IIE: comparative results for radiotherapy only, 69. Sasai K, Yamabe H, Haga H, et al. Non-Hodgkin’s lymphoma of combination chemotherapy only, and multimodality treatment. Int the thyroid. A clinical study of twenty-two cases. Acta Oncol 1996; J Radiat Oncol Biol Phys 1986;12:1807–12. 35:457–62. 78. Walt AJ, Woolner LB, Black BM. Primary malignant lymphoma 70. Sirota DK, Segal RL. Primary lymphomas of the thyroid gland. of the thyroid. Cancer 1957;10:663–77. JAMA 1979;242:1743–6. 79. Wiltshaw E. The natural history of extramedullary plasmacytoma 71. Skarsgard ED, Connors JM, Robins RE. A current analysis of and its relation to solitary myeloma of bone and myelomatosis. primary lymphoma of the thyroid. Arch Surg 1991;126:1199–203. Medicine (Baltimore) 1976;55:217–38. 72. Souhami L, Simpson WJ, Carruthers JS. Malignant lymphoma of 80. Woolner LB, McConahey WM, Beahrs OH, Black BM. Primary the thyroid gland. Int J Radiat Oncol Biol Phys 1980;6:1143–7. malignant lymphoma of the thyroid. Review of forty-six cases. Am 73. Staunton MD, Greening WP. Clinical diagnosis of thyroid cancer. J Surg 1966;111:502–23. BMJ 1973;4:532–5. 81. Wright CF, Reid AH, Tsai MM, et al. Detection of Epstein-Barr 74. Tennvall J, Cavallin-Stahl E, Akerman M. Primary localized non- virus sequences in Hodgkin’s disease by the polymerase chain Hodgkin’s lymphoma of the thyroid: a retrospective clinicopatho- reaction. Am J Pathol 1991;139:393–8. logical review. Eur J Surg Oncol 1987;13:297–302. 82. Yamaguchi M, Ohno T, Kita K. Gamma/delta T-cell lymphoma of 75. Tsang RW, Gospodarowicz MK, Sutcliffe SB, et al. Non- the thyroid gland [Letter]. NEnglJMed1997;336:1391–2.

Am J Surg Pathol, Vol. 24, No. 5, 2000