Expression of DBC1 and Androgen Receptor Predict Poor Prognosis in Diffuse Large B Cell Lymphoma¹

Patients and Samples

For this study, we used previously established tissue microarrays (TMAs) from 104 DLBCL patients diagnosed between December 1995 and April 2007 for whom the expressional status of CD10, Bcl6, MUM1, and Bcl2 was known [27]. Among the 104 cases included in the TMAs, three tissue cores had been exhausted by the previous study. For these three cases, we reviewed the original tissue blocks. However, there were no remaining usable samples. Therefore, we finally included 101 cases in the present study. This study had local ethics committee approval obtained from the Chonbuk National University Hospital's Institutional Review Board. Informed consent was provided according to the Declaration of Helsinki.

In the 101 patients included in the present study, 59 patients received cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) chemotherapy as an initial regimen. Since 2004, rituximab was added and 34 patients were treated with rituximab-CHOP (R-CHOP). Four patients were treated with combined chemotherapy with other regimens and four patients were treated with radiotherapy. The patients were grouped according to their age (<60 years vs ≥60 years), sex, stage (I and II vs III and IV), number of extranodal sites (<2 vs ≥2), serum lactate dehydrogenase (LDH) levels, Eastern Cooperative Oncology Group (ECOG) performance status (0–1 vs 2–4), presence of bulky disease, presence of B symptoms, presence of complete response (CR), immunohistochemical expression of Bcl2, and immunohistochemical subtype of DLBCL by the expression pattern of CD10, Bcl6, and MUM1 (GCB vs non-GCB) [28,29]. We have used the data for immunohistochemical expression for Bcl2 and immunohistochemical subtyping from our previously published report [27]. In addition, the patients were grouped by international prognostic index (IPI) scores [low (score of 0–2) vs high (score of 3–5)] [28,30]. Although there were no data for serum LDH levels in six patients, it was not influenced by the IPI categorization because the IPI scores without including serum LDH levels of six patients were zero, one, or three [27].

Immunohistochemical Staining and Scoring

Immunohistochemical staining was performed by using previously established TMAs. The antigen retrieval procedure was performed with a microwave oven for 12 minutes in sodium citrate buffer. The following markers were used: DBC1 (1:100; Bethyl Laboratories, Montgomery, TX) and AR (1:50; Santa Cruz Biotechnology, Santa Cruz, CA; clone N-20). Immunohistochemical scoring was performed by three pathologists (K.Y.J., S.J.N., and H.S.P.) under a multiviewing microscope with consensus without knowledge of the clinicopathologic information. Positivity for DBC1 and AR immunostaining was evaluated to estimate the percentages of positive tumor cells for each marker. Because it has been reported that DBC1 and AR are expressed primarily in the nuclei [6,7,23,24], we have evaluated only nuclear staining for the estimation of positivity for DBC1 and AR immunostaining. The cases showed nuclear immunoreactivity in more than 30% of tumor cells scored as positive according to the cutoff value used by others and us [6,7,27–29].

Statistical Analysis

End points of interest were overall survival (OS) and relapse-free survival (RFS). The follow-up end point was the date of last contact or the date of death through December 2011. The association between DBC1 and AR expression and other clinicopathologic factors were analyzed using the Pearson chi-squared test or Pearson correlation coefficient. OS was measured from the date of diagnosis to the date of death or last contact. Patients who were alive at last contact were treated as censored for OS analysis. RFS was measured from the date of diagnosis to the date of relapse, death, or last contact. Patients who were alive at last contact and who had not recurred were treated as censored for RFS analysis. Univariate and multivariate Cox proportional hazard regression analyses were performed for the survival analysis. The method by Kaplan and Meier was used to further illustrate the impact of OS and RFS where indicated. Five- and 10-year survival rates were also calculated. Statistical analyses were performed using SPSS software (version 18.0). P values less than .05 were considered statistically significant.

Association of DBC1 and AR Expression with Clinicopathologic Characteristics of DLBCL

The clinicopathologic features are summarized in Table 1. Responses to treatments were assessed according to international workshop criteria [31]. Among the 101 patients with assessable responses, 64 patients (63%) had CR and 37 patients (37%) failed to achieve CR. Nineteen of the 64 patients with CR (30%) relapsed, and 53 patients (52%) of the total group died. The median follow-up was 61.5 months (range, 1–191 months). The OS for the entire group of DLBCL patients was 48%. The 5- and 10-year survival rates for the entire group were 51% and 45%, respectively. By applying the TMA algorithm for the immunohistochemical staining for CD10, Bcl6, and MUM1 described by Hans et al., 37 (37%) patients were classified as GCB-type and 64 patients (63%) as non-GCB-type DLBCL [28]. The clinical features of the total DLBCL patients, GCB-type DLBCL, and non-GCB-type DLBCL are shown in Table 1.

As it has been previously reported, immunoreactivity of DBC1 in DLBCL tissues was found primarily in the nuclei [6,7]. Similarly, AR has been reported to be expressed primarily in the nuclei and minimally in the cytoplasm in hematopoietic malignant cell lines [23], AR expression was mainly localized in the nuclei of tumor cells in DLBCL, and weak cytoplasmic expression was also observed in some cases. The results of immunohistochemical staining for DBC1 and AR are shown in Figure 1. Positive expression of DBC1 and AR was seen in 73% (74/101) and 70% (71/101) of patients (Table 1). In total DLBCL patients, as shown in Table 1, DBC1 expression significantly correlated with advanced clinical stage (P = .027), elevated serum LDH level (P = .038), high IPI score (P = .017), and AR expression (P < .001). The correlation coefficient between DBC1 and AR expression was 0.782 (P < .001). The expression of AR was also significantly correlated with advanced clinical stage (P = .018), elevated serum LDH level (P = .015), high IPI score (P = .005), absence of CR (P = .024), and absence of Bcl6 expression (P = .041). In non-GCB-type DLBCL, DBC1 expression showed significant correlation with high ECOG performance status (P = .008), advanced clinical stage (P = .028), elevated serum LDH level (P = .010), and high IPI score (P = .007). AR expression also showed significant correlation with high ECOG performance status (P = .005), elevated serum LDH level (P = .010 and P = .029, respectively), and high IPI score (P = .004). However, there were no clinicopathologic factors that have statistically significant association with DBC1 or AR expression in GCB-type DLBCL. Of particular note, the correlation between DBC1 expression and AR expression was significant in both non-GCB-type (P < .001) and GCB-type (P < .001) DLBCL.

Expression of DBC1 and AR in DLBCL Was Correlated with Reduced RFS and OS according to Univariate Analysis

Univariate analysis of the expression of DBC1 and AR and its relationship to RFS and OS are shown in Table 2. For DLBCL in general, presence of B symptoms, high ECOG performance status, high clinical stage, increased serum LDH, high IPI score, absence of Bcl6 expression, and non-GCB-type DLBCL predicted shorter OS and RFS (Figure 2, A and B). DBC1 expression predicted shorter OS (P = .001) and RFS (P < .001; Figure 2C). Tumors expressing AR were associated with both significantly shorter OS (P = .003) and RFS (P = .002; Figure 2D). The 5- and 10-year survival rates of DBC1-negative patients were both 78%. In contrast, the patients with DBC1 expression had a 35% 5-year survival rate and a 28% 10-year survival rate. Similarly, the patients with AR expression had a 29% 10-year survival rate, and AR-negative patients had a 69% 10-year survival rate. The 5-year survival rate of R-CHOP-treated patients (65%) was longer than that of CHOP-treated patients (43%). Although, there was no difference in OS between R-CHOP- and CHOP-treated patients (P = .143), R-CHOP-treated patients showed favorable RFS compared with CHOP-treated patients (P = .024). In our DLBCL population, the follow-up duration was different between those two groups since rituximab was added in treatment regimens of DLBCL after 2004. The mean follow-up duration of the R-CHOP-treated group was 34 months (range, 1–92 months), while it was 59 months (range, 2–191 months) for the CHOP-treated group. Therefore, when we did survival analysis in the patients with less than 93 months of follow-up, R-CHOP-treated patients showed significantly longer OS (P < .001) and RFS (P < .001) than those of the CHOP-treated patients. In these patient groups, DBC1 expression also predicted shorter OS (P = .001) and RFS (P = .001) and AR expression also predicted shorter OS (P = .003) and RFS (P = .004).

In non-GCB type, high ECOG performance status, high clinical stage, high IPI score, DBC1 expression, and AR expression predicted shorter OS and RFS (Figure 3, A and B). In GCB type, high clinical stage, high number of extranodal involvement, and high IPI score predicted shorter OS. However, there was no factor significantly associated with RFS in GCB type. Only the expression of DBC1 was a possible predictor of RFS (P = .064; Figure 3, C and D). When separately considering DLBCL patients with low or high IPI scores, DBC1 expression predicted significantly worse OS and RFS within both low IPI (OS, P = .013; RFS, P = .023) and high IPI (OS, P = .036; RFS, P = .033) categories. AR expression also predicted significantly shorter OS (P = .028) within the low IPI category but not in high IPI category.

There was significant correlation between DBC1 and AR expression, and positive correlative roles between these two molecules in cell biology have been suggested. Therefore, we performed further survival analysis with the combined expression patterns of DBC1 and AR. As shown in Figure 4, DLBCL coexpressing DBC1 and AR (DBC1+/AR+ group) showed the most unfavorable OS and RFS compared with other subgroups in total DLBCL (log rank: OS, P = .001; RFS, P < .001) and non-GCB type (log rank: OS, P = .002; RFS, P < .001). The 10-year survival rate of the DBC1+/AR+ group was 26% in total DLBCL and only 17% in non-GCB type. In contrast, the 10-year survival rate of DBC1-/AR- group was 76% in total DLBCL and 75% in non-GCB type (Table 2).

Expression of DBC1 in DLBCL Was an Independent Prognostic Factor for Worse Survival Outcome by Multivariate Analysis

The variables significantly correlated with OS or RFS by univariate analysis were included in multivariate analysis for OS or RFS. The variables included in multivariate analysis for OS and RFS of total DLBCL were presence of B symptoms, ECOG performance status, presence of bulky disease, stage, serum LDH levels, IPI score, immunohistochemical subtype of DLBCL, and the expression of BCL6, DBC1, and AR. For total DLBCL, stage, ECOG performance status, presence of bulky disease, and DBC1 expression were independent prognostic factors significantly associated with both OS and RFS. The patients with DBC1 expression had a 3.137-fold [95% confidence interval (CI), 1.23–8.0] greater risk of lower OS (P = .017) and a 3.909-fold (95% CI, 1.536–9.95) greater risk of lower RFS (P = .004) than those who did not express DBC1. DBC1 expression was also a prognostic factor significantly associated with OS in non-GCB-type DLBCL. The patients having DBC1-expressing tumors had a 3.695-fold (95% CI, 1.098–12.437) greater risk of death than those who did not express DBC1 (P = .035). In non-GCB-type DLBCL, AR expression was significantly associated with lower RFS (P = .005) by multivariate analysis. However, in GCB-type DLBCL, only the clinical stage of DLBCL was an independent predictor of OS (P = .020). Other factors did not associate with OS or RFS by multivariate analysis in GCB-type DLBCL (Tables 3 and 4).