Prognostic Significance of CD20 Expression in Classical Hodgkin Lymphoma: A Clinicopathological Study of 119 Cases¹

HRS³ cells represent the neoplastic population of cHL. Recent immunological and molecular studies have shown that HRS cells originate from mature germinal center B cells that have undergone clonal immunoglobulin gene rearrangement, but have crippling mutations or other molecular alterations, preventing immunoglobulin transcription (1, 2, 3, 4, 5, 6, 7). Although cHL is genotypically considered a B-cell lymphoma, the classical B-cell marker CD20 is expressed only in ∼20–30% of such cases. Moreover, CD20 is the only marker of B-cell differentiation that can be repeatedly and reproducibly detected on HRS cells in paraffin sections (7, 8, 9). CD20 is a membrane-embedded, nonglycosylated, phosphorylated protein (10, 11) that appears on the B-cell surface after immunoglobulin light chain rearrangement and before expression of intact surface immunoglobulin (12). CD20 resembles a Ca²⁺ ion channel (11, 13) and is involved in signal transduction for B-cell differentiation and proliferation (14), as well as for G₀-G₁ cell cycle transition (15).

Clinically established prognostic factors in cHL are stage, age, presence of B-symptoms, bulky disease, hemoglobin concentration, erythrocyte sedimentation rate, and serum levels of albumin, lactate dehydrogenase, β₂-microglobulin, and interleukin-10 as well as association with HIV (16, 17, 18, 19, 20). The prognostic significance of CD20 expression in cHL is controversial and a matter of ongoing debate (19, 20, 21, 22, 23). The German Hodgkin Study Group reported that the CD30+/CD20+ immunophenotype of HRS cells has a positive but statistically insignificant impact on FFS in cHL (20). In the same study, however, CD30−/CD15−/CD20+ cHL had an inferior clinical outcome. In a recent investigation, Rassidakis et al. (19) found no association between CD20 status and FFS or OS in patients with stages I–IV cHL treated by at least seven efficacy-equivalent chemotherapy regimens. In line with these findings, Molot et al. (22) reported no prognostic role for CD20 expression in Hodgkin lymphoma, whereas the group from Memorial Sloan-Kettering Cancer Center (23) found an inferior clinical outcome in CD20-positive cHL patients.

To further assess the prognostic significance of CD20 expression in cHL, we performed a retrospective single institutional study of 119 cases with a mean clinical follow-up of 12 years.

We analyzed 119 formalin-fixed, paraffin-embedded samples from cHL cases from the files of the Department of Pathology at the University of Innsbruck, diagnosed between 1974 and 1999 (24). All cases were reclassified according to the updated WHO classification (8). Clinical data were obtained by reviewing the charts and contacting the treating physicians. Treatment was either standard or consistent with the investigational protocols active during the time the patients were diagnosed. In brief, patients were staged surgically in stages I and II when radiotherapy was administered as the sole treatment modality. Otherwise, patients were staged clinically and mostly treated with risk-adapted treatment strategies according to the protocols of the German Hodgkin Study Group. In advanced stages, radiotherapy was used only for treatment of residual disease or primary bulky disease (24). Disease remission was defined as absence of disease for at least 1 month as determined by laboratory and imaging studies as well as physical examination. Disease relapse was defined as disease progression occurring at least 1 month after achieving disease remission. Treatment failure was defined as disease relapse or primary treatment resistance.

The primary antibodies used in this study and types of antigen retrieval performed are listed in Table 1. Standard immunoperoxidase techniques were used with an automated immunostainer (Nexes; Ventana, Tucson, AZ) exactly as described previously (25). Diaminobenzidine was used as chromogen. For positive control staining, tonsils with follicular hyperplasia and with paracortical hyperplasia from patients with florid infectious mononucleosis were used. For negative controls, the primary antibodies were omitted. A case was considered CD20-positive if there was specific membranous staining in >10% of the HRS cells (Fig. 1). The staining intensity of small B lymphocytes within the cHL infiltrates served as internal control. Taking into account previous studies investigating the expression pattern of CD20 in cHL and its association with clinical outcome (19, 20, 21), we considered a cutoff value of 10% the most appropriate.

Statistical analysis was performed using the Statistical Package of Social Sciences (SPSS). FFS was analyzed with the Kaplan-Maier method and compared by the log-rank test. The Pearson χ² test was applied to demonstrate correlations between expression of CD20 and relapse rate as well as different clinical and laboratory parameters. To compare differences for patients treated before and after 1980, we used the independent sample t test. Multivariate analysis for the effect of expression of CD15 and CD20 as well as for age, sex, Ann Arbor stage, EBV association, B-symptoms, and histological subtype was performed using a general linear model. P < 0.05 was considered significant.

Our study group consisted of 65 (55%) male and 54 (45%) female patients. The mean ages were 40.4 years (range, 14–83 years) for the men and 34.3 years (range, 16–77 years) for the women. Eighty-six of the 119 patients (72%) were younger than 45 years of age. Histopathologically, this group included 70 cases with nodular sclerosis, 39 with mixed cellularity, 7 with lymphocyte-rich cHL, and 3 with lymphocyte-depleted cHL. According to the Ann Arbor classification, 12 cases presented in clinical stage I, 56 in stage II, 27 in stage III, and 15 in stage IV; 31 cases (26%) were associated with EBV, as assessed by the expression of LMP-1. Fifty-four patients were treated with combined radiochemotherapy, 30 were treated with chemotherapy, and 24 were treated with radiotherapy alone. Of the 84 patients treated with combined radiochemotherapy or chemotherapy alone, 18 received ABVD + COPP, 24 received BEACOPP, 27 received ABVD, and 15 received COPP chemotherapy regimens (Table 2). Within the median follow-up period of 127 months (range, 0.2–331 months), OS was 59%, and the average time to first relapse was 52 months (range, 5.7–186 months). Disease relapses were observed in 32 patients (27%). During the course of our study, there were seven malignancy-related deaths, six attributable to disease relapse and one attributable to primary refractory disease. In addition, 11 patients died from cardiovascular and pulmonary diseases, 6 from the development of a second malignancy, and 1 from trauma. During the course of this study, FFS in the whole collective increased from 45% for patients treated from 1974 to 1980 to 65% for patients treated from 1981 to 1999 (P = 0.007).

All cHLs in our study group were CD30-positive, and 113 of 119 (95%) were also positive for CD15. CD20 was expressed in 20% (24 of 119) of the cases (Table 2). With regard to therapy regimens, CD20 was expressed in 9 of 54 (17%) patients treated with combined radiochemotherapy, in 7 of 30 (23%) treated by chemotherapy, and in 6 of 24 (25%) treated by radiotherapy. The clinicopathological patient characteristics of CD20 expression are summarized in Table 2.

CD20 expression was asymmetrically distributed in the age groups (15% CD20-positive cases in patients <45 years of age, 33% in patients >45 years; P = 0.027). There was no correlation of CD20 expression with EBV infection of HRS cells (P = 0.7). Univariate analysis revealed a significantly higher frequency of disease relapses in the CD20-negative group (30 of 95; 32%) compared with the CD20-positive group (2 of 24; 8%; P = 0.022). Compared by the log-rank test, the FFS was worse in the CD20-negative than the CD20-positive group; P = 0.0195; Fig. 2).

We then evaluated relapse rates in CD20-positive and -negative cases within the different patient groups, using as criteria treatment regimens, including the different types of chemotherapy; age; sex; disease stage; histological subtypes; LMP-1; and CD15-positive and -negative clusters as well as the presence or absence of B-symptoms. When we applied the log-rank test, CD20 expression appeared to be a statistically significant prognostic factor for FFS in all groups except for age and stage, where it was of borderline significance (P = 0.052 and 0.057, respectively; Table 3). For the different types of chemotherapy, the unequivocal trend for a better FFS of CD20-positive patients remained; however, because there were too few cases in each group, this appeared not to be statistically significant. Multivariate analysis revealed that therapy regimens (P = 0.013), different types of chemotherapy applied (P = 0.011, with patients receiving BEACOPP showing the best and ABVD the worst results), B-symptoms (P = 0.014), and age (P = 0.042) were independent prognostic factors for OS and that therapy regimens (P = 0.0005) and stage (P = 0.013) were independent prognostic factors for FFS. CD20 did not appear to be an independent positive prognostic factor for FFS or OS in this cHL study group, but the P for FFS was of borderline significance (P = 0.058). Finally, malignancy-related mortality was exclusive to CD20-negative cases (n = 7), although this was not statistically significant.

When analyzed separately for patients treated in the period 1974–1980, CD20 expression appeared to be of independent prognostic significance for both OS (P = 0.004) and FFS (P = 0.035). The prognostic significance of CD20 disappeared in the period 1981–1999 because of better FFS in the CD20-negative group (75% for 1981–1999, compared with 43% for 1974–1980; P = 0.005; Table 3). Comparisons with the t test revealed no significant differences in disease stages, B-symptoms, age, EBV association, and CD20 status between patients treated in the period 1974–1980 and those treated during 1981–1999, except for the specific therapy regimens applied (P = 0.003). After 1980, the proportion of patients treated by radiotherapy alone was low (25% for 1981–1999, compared with 41% for 1974–1980), but the proportion of patients treated by combined radiochemotherapy or chemotherapy who received ABVD + COPP or BEACOPP chemotherapy regimens was high.

In this single institutional study of 119 cases of CD30-positive cHL with a mean follow-up period of 12 years, we demonstrate that the disease immunophenotype CD30+/CD15±/CD20+ (20) is associated with a favorable clinical outcome. This was true for all histological subtypes and Ann Arbor disease stages, CD15± and LMP-1± cases, male and female patients from both age groups, patients with and without B-symptoms, and for all treatment regimens (Table 3). In a multivariate analysis of all these factors, CD20 was not an independent prognostic factor for FFS. However, the P was of borderline significance (P = 0.058), indicating that analysis of a larger patient population is warranted and will likely produce significant conclusions.

Interestingly, the prognostic value of CD20 expression for FFS in our series, although significant for the treatment period 1974–1980, disappeared in the period 1981–1999. van Spronsen et al. (26) made the same observation for the prognostic significance of histopathological grading in nodular sclerosing cHL. Thus, for the first time in cHL, we demonstrate that the significance of an immunohistochemically determined prognostic marker can be neutralized by improved therapy regimens.

Our results are at variance with the few studies investigating the association of CD20 expression with clinical outcome in cHL (19, 20, 21, 22, 23). The German Hodgkin Study Group reported an inferior FFS and OS in 21 patients of 1286 who were classified as cHL but expressed only CD20 (CD30−/CD15−/CD20+ cHL; Ref. 20); however, it is not clear whether these cases may in part represent other entities, such as nodular LPHL or T-cell-rich large B-cell lymphomas, or borderline cases, setting them apart from true cHL. The German Hodgkin Study Group also indicated that these tumors might represent lymphomas other than cHL and recommended histopathological reevaluation in these cases, the results of which have not yet been reported. Interestingly, in the same study, CD30+/CD15±/CD20+ cHL cases had a better FFS than did CD30+/CD15±/CD20− cases, in agreement with the results of our study. These differences did not appear to be statistically significant (20), but the mean follow-up period was shorter than in our study.

Donnelly et al. (23) from the Memorial Sloan-Kettering Cancer Center described an association of CD20 expression in cHL with worse clinical outcome. Unfortunately, their data have been published only as an abstract, and it is not clear what cutoff value was used to define CD20 positivity. CD20 was detected in only 9% (16 of 181) of cHL cases, and the follow-up period was considerably shorter than in our study.

Recently, Rassidakis et al. (19) analyzed 598 previously untreated cHL patients for the prognostic significance of CD20 expression. CD20 was expressed by HRS cells in 22% of patients but was not associated with different FFS after treatment with equivalent regimens. The discrepancy with our findings might be explained by differences in the follow-up period. The mean follow-up period in the study of Rassidakis et al. (19) was 65 months versus 144 months in our study, and we observed 11 late relapses that occurred after follow-up month 65 in the CD20-negative group.

Finally, in the study of Molot et al. (22), who reported no effect of CD20 expression on clinical outcome, only patients with stage I and II disease treated solely with radiotherapy were included, and no exact definition of CD20 positivity was given. Five of their nine CD20-positive cases were classified as LPHL, which is at present considered a distinct clinicopathological entity differing from cHL in clinical features and behavior as well as morphology and immunophenotype (8). Interestingly, none of the remaining four patients with CD20-positive cHL have relapsed or died (22).

In our study, 20% of the cHL cases (24 of 119) expressed CD20 based on a cutoff of 10%, in agreement with most previous studies [Refs. 8 (and references therein), 9]. However, according to the literature, the frequency of CD20 expression in cHL can range from 5 to 80% (7, 8, 19, 20, 21, 22, 23). This broad range is explained by different fixation procedures and staining techniques, use of different cutoff levels, and at least in a small proportion of cases, inclusion of other, morphologically closely related entities, such as LPHL and T-cell-rich large B-cell lymphoma (27, 28), which are characterized by different clinical behavior. Interestingly, Rassidakis et al. (19) showed that even when various cutoffs were used to define CD20 positivity in HRS cells (>0, 5, 10, 20, 30, and 50%), there was no significant influence on FFS, although the percentage of CD20-positive cHL declined. Nonetheless, a uniform definition of CD20 positivity in HRS cells is needed before the clinical potential of these results can be realized in the treatment of cHL with the anti-CD20 monoclonal antibody rituximab (29). This would allow a direct comparison of the data from different clinical trials (30). Moreover, it would also address the question of whether there is a correlation between expression of CD20 in HRS cells and response to anti-CD20 therapy in cHL. Interestingly, recent in vivo and in vitro studies of non-Hodgkin lymphoma demonstrated that the quantitative expression of CD20 molecules on distinct histological types of lymphoma correlates with rituximab efficacy (31, 32). In this context, it is also noteworthy that rituximab response rates up to 100% have been reported in patients with LPHL, who express the CD20 antigen in a high density (30, 33, 34, 35).

The pathophysiological mechanisms of the positive effect of CD20 expression on FFS in cHL are obscure. As mentioned previously, CD20 resembles a Ca²⁺ ion channel (11, 13). An increase in Ca²⁺ permeability in HRS cells, along with chemo- and/or radiotherapy, might decrease apoptotic resistance or even activate programmed cell death, as it does in other lymphomas (29). Furthermore, CD20 can inhibit some effects of interleukin-4 (4), a potent autocrine growth factor for HRS cells (36). On the other hand, it might be possible that expression of CD20 on HRS cells is merely a marker of a different gene expression program associated with favorable clinical outcome. Whether this is true remains to be determined in future studies. At least in our group, there was no association between EBV infection and expression of CD20.

We thank A. Zimpfer for excellent technical assistance.