INTRODUCTION

The skin is the second most common site for extranodal lymphomas, after the gastrointestinal tract (1). Although follicle center lymphoma (FCL) most often occurs in lymph nodes, it may also present in the skin as both a primary or secondary manifestation of disease and is the most common B-cell lymphoma of the skin (2, 3). A lymphoma is considered to be “primary” to the skin if there is no extra-cutaneous disease at the time (and within the first 6 months) of its diagnosis (2, 4, 5). Primary cutaneous FCL has been reported in the European Organization for Research and Treatment of Cancer (EORTC) for Cutaneous Lymphomas (2) and in other studies as a distinct entity, which has a good prognosis and rarely disseminates to extra-cutaneous sites (4, 6, 7). But it has been argued, that the entity classified as cutaneous FCL in the EORTC is not the same as those described in other classifications (5, 8). It is uncertain if the World Health Organization Classification of Neoplastic Diseases of Lymphoid Tissues (WHO) will consider the entity distinct, although in one proposed schema it has been proposed as a separate entity (9). Other suggestions have included cutaneous FCL as a variant of FCL (10). The Revised European-American Classification of Lymphoid Neoplasms (R.E.A.L.) does not recognize cutaneous FCL as a distinct entity but includes it in the broader classification of FCL (11). Sander et al. (12, 13) have proposed that the R.E.A.L. be used for cutaneous FCL but state that cutaneous FCL appears to be distinctive. Because the R.E.A.L. is the most current classification used in the United States by hematopathologists, is adequate for diagnosing cutaneous lymphomas, and the grading of FCL in this classification is well established, we classified our cases according to the R.E.A.L. classification. Using the R.E.A.L. classification is also in line with the proposed view that organ specific classifications are not necessary for recognition of common features of lymphoma involving multiple sites (13, 14). In our study, we investigated the clinical, histologic, immunophenotypic and molecular genetic features of cutaneous follicle center lymphomas. We included CD10 (15, 16), bcl-6 (marker of follicle center cells) (17), bcl-2 (18), t (14;18) (19), and trisomy 3 (20) in an attempt to better characterize cutaneous follicle center lymphomas and to separate them from other small B-cell lymphomas that can appear follicular (21).

MATERIALS AND METHODS

Case Selection

Nineteen cases diagnosed as cutaneous “follicular,” “follicle center,” or “nodular” lymphoma without extra-cutaneous involvement were selected from the files of the Dermatopathology Registry at the Armed Forces Institute of Pathology from 1990 to 1997. There were hematoxylin-eosin stained slides and paraffin blocks (or unstained slides) in all cases. All tissues were formalin fixed. Nine cases had more than one biopsy performed. Follow- up was obtained in 17 cases. All cases were classified and graded according to the Revised European-American Classification of Lymphoid Neoplasms (11).

Immunohistochemistry

The immunohistochemical antibodies are shown in Table 1. Five micron sections from paraffin embedded tissue blocks were prepared for immunophenotypic analysis according to the standard avidin-biotin complex method of Hsu et al. (22). A complete antibody panel for CD45RB, CD20, CD3, CD5, CD43, CD10, bcl-6 and bcl-2 was performed in 14/19 cases in at least one specimen if more than one was available. A partial panel was performed in 5/19, based on availability of material. Cyclin D1 (bcl-1) was performed in two cases, one CD5 positive and one with equivocal CD5 reactivity.

TABLE 1 Antibodies
Full size table

Molecular Diagnostic Studies

Molecular diagnostic studies were performed by polymerase chain reaction (PCR) from formalin-fixed paraffin-embedded tissue in 18/19 cases in at least one biopsy if multiple biopsies were available. Studies included immunoglobulin heavy chain (IgH), T-cell receptor (TCR) β, and t (14;18) to detect the major breakpoint region.

IgH gene rearrangement was detected by a PCR method using published consensus primers (FR3A, V region and CFW1 J region) (23). TCR β was detected by a PCR published PCR method using four sets of primer pairs (VJ1, VJ2 D1J2, D2J2) as previously published (23). TCRβ has been described in B-cell lymphoma it therefore would not preclude the diagnosis of a B-cell lymphoma but may indicate malignancy (24). The major breakpoint region of t(14;18) was detected by primers (CH18-515) and probes (pMBR) that have been previously published (25). Her2/neu was used as a control to test for DNA integrity.

Fluorescent In situ Hybridization for Trisomy 3

The trisomy 3 assay was performed in 13/19 cases. A deparaffinization kit was used according to the manufacturer's instructions (Vysis Inc, Dowers Grove, IL.) for trisomy 3 analysis. The DNA probes for chromosome 3, region 3p11.1-q11.1 (Vysis) and chromosome 16, region 16q11.2 (Vysis) were labeled and analyzed using the manufacturer's procedure for CEP Chromosome Enumeration DNA FISH Probes (Vysis). The in situ signals were demonstrated by fluorescence microscopy. Signals were analyzed for the percentage of cells with 0, 1, 2, 3, and 4 signals. Twelve percent or higher for three signals were considered positive for trisomy 3. The slides were counterstained with DAPI II counterstain (Vysis). Chromosome 16 was used as a control. The assays were performed in triplicate. Twelve percent was used as a cutoff for the lower limit of normal. We used the following to arrive at 12% as a reasonable cutoff: the manufacturer's recommendation of 10 to 11% and previously published data showing that marginal zone lymphomas have a mean of 16 to 46% positivity for trisomy 3 by FISH with 7 to 15% as the lower level of positivity for marginal zone lymphoma (20, 26, 27, 28).

RESULTS

Clinical Features

The clinical results are summarized in Table 2. There was a near equal distribution of males (9) and females (10) with ages ranging from 33 to 88 years (mean, 64 years). All of the 19 patients presented with cutaneous lesions with no other sites of involvement. Sites of disease included head and neck (12/19) and back or trunk (6/19) and arm (1/19). In 2 patients, no clinical staging history was available after the initial biopsy.

TABLE 2 Clinical Features
Full size table

Histology

All cases showed a nodular or vaguely nodular, dense atypical lymphoid infiltrate in the deep dermis. There was a grenz zone, or sparing of the epidermis, in 18/18 cases. (In one case, the epidermis was not present for examination). Four of 19 cases were Grade I/III (small cell), 5/19 were Grade II/III (mixed small and large cell) and 8/19 were Grade III/III (large cell), 2/19 had a first biopsy graded as II/III (mixed small and large cell) and a second biopsy graded as III/III (large cell). The mantle zones were well preserved in 4/19, attenuated in 9/19 and absent in 6/19. One case (case 4) showed minimal marginal zone differentiation. In all cases there were intervening centrocytes mixed with the centroblasts with the exception of one case comprised of exclusively centroblasts. In one other case (case 15), the nodules were predominately composed of centrocytes without intervening centroblasts. Tingible body macrophages were seen in only one case. All cases lacked polarity of the malignant follicles (Fig. 1).

FIGURE 1
figure 1

Case 14. Follicle center lymphoma, Grade III/III. A, deep dermal involvement with a nodular pattern. B, malignant nodule with attenuation of the mantle zone. C, large, atypical centroblasts and centrocytes. D, immunoreactivity for CD10 (left) and bcl-6 (nuclear; right).

Full size image

Immunohistochemistry

The morphologic, immunohistochemical and molecular genetic results are shown in Table 3. All cases were reactive with CD20 and negative with CD3. Bcl-2 was reactive in 11/18 cases. Bcl-6 was immunoreactive in 15/15 cases. CD10 was immunoreactive in 14/17. CD43 was immunoreactive in 2/16 cases. Both of the CD43 immunoreactive cases were also CD10 and bcl-6 immunoreactive. One of 17 cases were immunoreactive for CD5 and one case was equivocal; the CD5 immunoreactive case was bcl-6 positive and CD10 negative; both were negative with bcl-1. Eleven of 13 cases showed an organized dendritic network in the malignant nodules with CD21.

TABLE 3 Histology, Immunophenotype and Genotype
Full size table

Molecular Genetic Studies

Eight of 18 cases studied showed monoclonal bands for immunoglobulin heavy chain (IgH) (one case was non-amplifiable and 9 were negative). Three of 17 showed the presence of t(14;18), cases 2, 5 and 10 (one case was technically unsatisfactory and one had no tissue to perform the assay). One case showed both monoclonal IgH and t(14;18). All negative cases for t(14;18) showed satisfactory DNA integrity with Her2 neu. When TCR testing was performed it is indicated in Table 3. One specimen showed TCRβ monoclonality but also showed t(14;18).

Fluorescence In Situ Hybridization

The immunohistochemical, molecular and FISH results are summarized in Table 3. Four of 13 cases tested showed the presence of trisomy 3 above the threshold. Of the 4 cases with trisomy 3, none showed the presence of t(14;18). Two of the 4 cases with trisomy 3 showed immunoreactivity with bcl-6 and CD10, which strongly supports follicle center origin (and not extranodal marginal zone B-cell lymphoma). One showed immunoreactivity with bcl-6 but was negative with CD10. The remaining case was only examined with molecular studies and did not show t(14;18) or IgH rearrangements. Case 9 is shown in Figure 2 and was interpreted as positive for trisomy 3 (16%). No cases were positive for trisomy 16, a control marker.

FIGURE 2
figure 2

Case 9A. Follicle center lymphoma, Grade II/III with 16% trisomy 3. A, deep dermal nodule showing an absent mantle zone. B, a mixture of small cleaved and large cells.

Full size image

Follow-Up

Overall, 14 of 17 patients are alive with no evidence of disease (ANED) from 4 to 137 months. Three of 17 are alive with disease (AWD) from 17 to 100 months. No follow- up information could be obtained in 2 cases. When examined by cytologic subtypes, follow-up shows that of patients with Grade I/III, 4/4 were ANED. Of the patients with FCL Grade II/III, 2/4 were ANED, 2/4 were AWD and in 1/4 no FU was obtained. Of patients with Grade III/III disease, 6/8 were ANED, 1/8 AWD, and in 1/8 no FU could be obtained. Two patients had 2 biopsies the first showing Grade II/III and the second showing Grade III/III. One of these patients was AWD and the other ANED.

DISCUSSION

Cutaneous follicle center lymphoma is considered by the EORTC classification to be a distinct clinicopathologic entity from nodal FCL. The R.E.A.L. does not recognize cutaneous FCL as a separate entity but considers it to be an extranodal presentation of FCL. The R.E.A.L. does not address the specific clinical behavior of cutaneous FCL but includes it within the broader FCL, which is “generally indolent” (11). The EORTC classification indicates that cutaneous FCL has indolent clinical behavior with a 5-year survival of 97%. It occurs most often in the head and neck region with only rare dissemination to extra-cutaneous sites and does not appear to have shorter survival with higher-grade lesions (2, 23). In contrast, the reported overall 5-year survival of nodal FCL is 67%, although higher grade FCL (follicular large cell versus follicular small cell) offered worse survival (29). But there are exceptions, with occasional cases of nodal FCL having long survival with little or no treatment even in those cases having large cell morphology (30). Our findings support the reported indolent clinical behavior and show no correlation in survival with grading of the lymphoma.

Survival in noncutaneous FCL also has been reported to have worse prognosis when bcl-2 rearrangements are absent, a finding not typically seen in cutaneous FCL (31). We report a low level of t(14;18) positivity in our cases by PCR (3/17 or 18%). This low percentage of t(14;18) may be inherent in the PCR method from paraffin embedded tissue, rather than a true lack of t(14;18) because there have been similar low levels of t(14;18) in previously reported findings in nodal FCL of 20 to 50% in paraffin embedded tissues (32, 33, 34, 35). In contrast to our study, a recent study reported no cases of primary cutaneous follicle center lymphoma possessed the t(14;18) (4). Although we feel that our findings reflect true presence of t(14;18), even in the present study, the rate of t(14;18) in cutaneous FCL is much less than nodal FCL, which in our laboratory is approximately 40% using PCR from paraffin tissue. Bcl-2 protein expression does not appear to reflect the incidence of t(14;18). However, bcl-2 protein expression was higher in the present study, than that previously reported in which bcl-2 was positive in only 0 to 46% (4, 33, 36, 37, 38). This also includes the EORTC, which notes that follicular lymphomas rarely express bcl-2 (2). A recent study found 3 of 4 cases showed immunoreactivity for bcl-2 but these cases were secondary (39). Because bcl-2 has been reported in a variety of small B-cell lymphomas without t(14;18) it is less specific than once thought in FCL, both cutaneous and in lymph node.

Bcl-6 and CD10 are both markers of follicle center cells and are immunoreactive in noncutaneous follicle center lymphomas (16, 40, 41). Our study shows they are also immunoreactive in the cutaneous form of FCL. Previous reports are not consistent in their finding of CD10 immunoreactivity. Reports by Willemze et al. (2), Berti et al. (42), and Santucci et al. (43) show no CD10 immunoreactivity, although a report by Garcia et al. did show cases that were immunoreactive with CD10 (44), and Yang et al. showed 14/15 cases immunoreactive (33). CD10 immunoreactivity is in keeping with Norton's (8) view that the true FCL of the skin have a similar immunophenotype to nodal FCL. Bcl-6 is a recently recognized germinal center cell marker (40), which has been reported positive in 15/15 cutaneous FCL in one previous study by Yang et al. (33). We found these two markers particularly helpful in differentiating FCL from both mantle cell lymphoma and marginal zone B-cell lymphoma, which support previous reported findings in nodal disease (17). We also found that CD21 highlighted the organized follicular dendritic cells in the malignant nodules, particularly in the lower grades, and was also helpful; a finding similar to that in nodal FCL (11). These findings also are supportive of the view that an organ specific classification such as the EORTC is not required and indeed may impede the recognition of common features of lymphomas involving multiple anatomic sites (13, 14).

The significance of CD5 or CD43 immunoreactivity in cutaneous follicle center lymphoma is unknown, but our study suggests that this co-expression with CD10 does not adversely affect clinical outcome. The low expression of CD43 in our study is consistent with previous reports of rare cases of FCL (45, 46). The two CD43 immunoreactive cases in this study are Grade III. Lai et al. also reported Grade III FCL expressing CD43 more often than Grade I or II (45). The coexpression of CD5 and CD10 has also been reported other small B-cell lymphomas (47) and in a recent report by Tiesinga et al. describing a unique floral variant of FCL (48). We do not know the significance of the CD5 expression in this study.

The percentage of cells determined with FISH to be positive for trisomy 3 (above the 12% threshold) in our study is 16 to 22% (mean 19%) and is in the scope of previously reported mean values in marginal zone B-cell lymphoma of the gastrointestinal tract (16 to 46%). Of note, the actual number of positive cells in previous reports varied greatly (7.5 to 86%) (20, 26, 27, 28). Blanco et al. and Wotherspoon et al. also examined other types of lymphoma and found low levels of trisomy three (20, 26). Our findings likely represent a true low level of trisomy 3, similar to previous reports in which +3 and +3p have been found in FCL in lymph node (49, 50). When seen in nodal FCL it is generally in Grade II or III, which may or may not also have the t(14;18) (49, 50, 51). Trisomy 3 in high-grade lymphoma may impart a good prognosis (49). None of our cases with trisomy 3 has marginal zone differentiation (marginal zone differentiation was minimally present in only one of our cases of FCL without trisomy 3); therefore it is not a likely explanation for the trisomy 3 positivity by FISH. Cytogenetics performed on fresh tissue, which is not available to us, would be helpful in further characterizing this finding.

The findings in this study highlight the morphologic and immunophenotypic and genotypic features of cutaneous FCL. Previous studies of cutaneous FCL may have included other small B-cell lymphomas with a nodular appearance making it difficult to “sort out” the reported findings. With the new paraffin markers available, it is more likely that cutaneous follicle center lymphoma will become a better-defined entity. Cutaneous FCL, it appears, has an immunophenotype similar to that of nodal FCL, although it may act in a more indolent manner, but this remains to be substantiated. Continuing study is necessary to completely characterize this neoplasm.