Clinical Relevance of the Lung Resistance Protein in Diffuse Large B-Cell Lymphomas¹ Free

DLBCL³ can be effectively treated with conventional polychemotherapy regimens with or without radiotherapy (1, 2). In addition, high-risk patients may benefit from either high-dose consolidation treatment with hematopoietic stem cell support after having achieved complete response from initial chemotherapy (3) or initial high-dose induction chemotherapy with stem cell support (4). Despite these improvements, 40–50% of the patients are not cured by chemotherapy because of drug-resistant disease.

MDR is one important type of drug resistance that is clinically relevant in leukemias (5, 6) and several solid tumors (7). Different mechanisms can contribute to MDR; some of them have already been studied in non-Hodgkin’s lymphomas. MDR1/P-glycoprotein expression occurs with various frequencies in lymphomas and is associated with clinical drug resistance to various anticancer drugs including anthracyclines and Vincaalkaloids (7, 8). Clinical trials to overcome P-glycoprotein-mediated resistance in drug-refractory lymphoma by combining chemotherapy with resistance modifiers indicated that, at least in a subset of patients with drug-refractory lymphoma, modulation of P-glycoprotein function is feasible, which suggests that P-glycoprotein expression plays a role in the drug resistance of this disease (9, 10, 11). MRP1, another important factor involved in MDR, is also expressed in lymphomas (12), but its impact on clinical outcome remains to be determined. Alterations in apoptosis and cell cycle regulation are also involved in drug resistance of lymphomas (13, 14, 15, 16). p53mutations were associated with poor outcome of chemotherapy and shorter survival in aggressive B-cell lymphomas (13) and in relapsed or drug-refractory non-Hodgkin’s lymphomas (14). The cyclin-dependent kinase inhibitor p27^Kip1 has also been shown to be involved in drug resistance (17),and lack of its expression is associated with shorter disease-free survival and overall survival in patients with DLBCL (16).

LRP is another protein that is associated with MDR. It was first detected in a non-P-glycoprotein-multidrug-resistant lung cancer cell line (18) and has been shown to be the human major vault protein (19). Vaults are complex ribonucleoprotein particles that, in addition to the major vault protein, also contain several minor vault proteins and a small RNA (20). Vaults are located mainly in the cytoplasm and, to a smaller degree, also in the nuclear membrane. They are believed to mediate intracellular and,in particular, nucleocytoplasmic transport (20). LRP expression of tumor cell lines is associated with resistance to doxorubicin, vincristine, carboplatin, cisplatin, and melphalan (21). A recent report by Kitazono et al.(22) provides evidence that LRP expression is involved in resistance to doxorubicin, vincristine, etoposide, paclitaxel, and gramicidin D and that LRP is associated with the transport of doxorubicin from the nucleus to the cytoplasm. LRP is physiologically overexpressed in colon tissue, lung tissue, renal proximal tubules,adrenal cortex and macrophages, but its physiological function remains to be evaluated (23).

Because LRP and MRP1 affect drugs commonly used in the treatment of DLBCL, expression of these proteins may affect response to chemotherapy and survival in DLBCL. To further address this possibility, we have studied LRP expression and MRP1 expression in lymphoma cells and their association with both response to chemotherapy and survival of the patients.

Forty-eight previously untreated patients (21 females, 27 males) with DLBCL, diagnosed between 1991 and 1999, were included in this study. All of the biopsy samples were diagnosed at the Institute of Clinical Pathology, University of Vienna, Vienna, Austria. Between 1991 and 1994, diagnosis was based on the criteria of the Kiel classification (24). Later on, lymphomas were typed according to the criteria provided in the Revised European-American Lymphoma (REAL) classification (25, 26, 27), and the former cases were reviewed and also classified according to REAL criteria. Lymphomas were subtyped using standard histological and immunohistological methods. Cases with antecedent low-grade B-cell lymphomas were not included in this study.

The clinical characteristics of the patients are summarized in Table 1. All of the patients received polychemotherapy. Twenty-eight patients were treated with CHOP, 19 patients with ProMACE-CytaBOM, and 1 patient with CEP. Patients with stage I disease received 3–4 cycles of chemotherapy plus consecutive radiotherapy (1). Patients with stage II-IV disease received 6 cycles of chemotherapy. Two of them with bulky disease who achieved a complete response were treated with additional radiotherapy after 6 cycles of chemotherapy. All of the patients were evaluable for response. Response to chemotherapy was assessed according to standard criteria (28). Complete response was defined as the absence of clinical and radiological evidence of disease for a minimum of at least 2 months. Eight patients received additional autologous bone marrow transplantation. In these patients, response assessment was performed after chemotherapy but before the transplantation.

Age, tumor stage, serum lactate dehydrogenase, performance status, and the number of extranodal sites of the disease were used to determine the International Prognostic Index (29). For statistical analysis, patients were grouped into low-risk (International Prognostic Index, 0–1), intermediate-risk (International Prognostic Index, 2–3),and high-risk (International Prognostic Index, 4–5) patients.

Immunohistochemistry was performed on formalin-fixed, paraffin-embedded lymphoma specimens. Paraffin sections were mounted on poly-l-lysin-coated glass microslides.

Sections were deparaffinized and rehydrated by consecutive submersions in xylene (two changes, 10 min each), absolute ethanol (two changes, 5 min each), 70% ethanol (two changes, 5 min each), and distilled water(3 min). Endogenous peroxidase activity was blocked by incubation in 0.06% H₂O₂ for 10 min at room temperature, and slides were washed in PBS. The tissues were preincubated for 20 min in normal serum (normal goat serum 1:50; Dako,Glostrup, Denmark) prior to an incubation for 2 h with either the LRP-56 monoclonal antibody (Alexis, Läufelfingen, Switzerland) or the MRPr1 monoclonal antibody (Alexis). Antibody binding was detected by the avidin-biotin-peroxidase method. Bound peroxidase was developed with 3,3′-diaminobenzidine (Dako). The slides were counterstained with Mayer’s Hämalaun and mounted with Aquatex (Merck, Darmstadt,Germany). All of the washes were performed in PBS.

Negative controls were performed as described above but without the monoclonal antibody. In some cases, additional controls with an irrelevant isotype-specific antibody (IgG2b for LRP-56 and IgG2a for MRPr1) were done. There was no difference in staining between the irrelevant isotype-specific antibodies and the negative controls without any primary antibody. Normal human kidney tissue, which is known to overexpress LRP and MRP1 (23, 30), was used as positive control.

Staining of lymphoma cells was examined by two investigators (M. F. and I. S.) without prior knowledge of the clinical outcome of the patients. Specimens were scored for the percentage of lymphoma cells showing granular cytoplasmic staining in the case of LRP and cytoplasmic and/or membrane staining in the case of MRP1. Staining was compared with corresponding negative and positive controls.

Associations between LRP or MRP1and clinical as well as laboratory parameters were assessed by χ² test, Fisher‘s exact test, or exact Mann-Whitney test. Survival probabilities were calculated with the product limit method according to Kaplan-Meier (31). Overall survival time was defined as the period between the time of diagnosis and the time of death. Survival times of patients still alive or patients who underwent bone marrow transplantation were censored with the date of the last follow-up or transplantation, respectively. Differences between survival curves were analyzed by means of the log-rank test. Logistic regression models and Cox proportional hazards regression models were used to assess the independent effects of covariables on survival (32). All of the Ps are results of two-sided tests. The SAS statistical software system (SAS Institute Inc., Cary, NC, 1990) was used for calculations.

LRP expression of previously untreated patients was immunohistochemically determined by means of the monoclonal antibody LRP-56. LRP staining was detected as characteristic granular cytoplasmic staining and ranged from 0 to 60%. In the case of positive staining, at least 10% of the lymphoma cells were stained. LRP expression was scored positive if any of the lymphoma cells showed brown cytoplasmic staining. LRP was positive in 11 (23%) of 48 lymphoma specimens at diagnosis (Table 1).

MRP1 expression was determined by means of the MRPr1 antibody. MRP1 staining ranged from 0 to 30% of the lymphoma cells. Any brown staining of lymphoma cells either cytoplasmic or membranous was scored as positive expression; MRP1 expression was positive in 21 (44%) of 48 samples (Table 2).

Next, we addressed the question as to whether LRP or MRP1 expression correlated with clinical or laboratory parameters. The major clinical and laboratory findings of the patients are summarized in Table 1. There was no significant association between LRP expression and age,sex, and β₂-microglobulin (Table 1). However,LRP expression was more frequently observed in patients with stage III and IV disease and in patients with elevated serum lactate dehydrogenase (>240 units/liter; Table 1). Whereas 24 of 25 low-risk patients (International Prognostic Index, 0–1) were LRP-negative, 6 of 18 intermediate-risk (International Prognostic Index, 2–3) and 4 of 5 high-risk patients (International Prognostic Index, 4–5) were LRP-positive (P = 0.0001; Table 1).

MRP1 expression was independent of age, sex,β ₂-microglobulin, serum lactate dehydrogenase,and the International Prognostic Index (Table 2). In addition, no correlation between MRP1 and LRP expression was observed (data not shown).

All of the 48 patients received polychemotherapy. The treatment protocols were equally distributed among LRP-negative and LRP-positive patients (Table 1). All of the patients were evaluable for response to chemotherapy. The complete response rate of the total study population was 54%. Partial responses and no responses were seen in 21 and 25%of the patients, respectively. The complete response rate was 65% for patients without LRP expression but only 18% for patients with LRP expression (P = 0.006; Table 3). Partial responses and no responses occurred in 8 (22%) and 5 (13%) of LRP-negative patients but in 2(18%) and 7 (64%) of LRP-positive patients (data not shown). With regard to MRP1, the complete response rate was 56% for MRP1-negative patients and 52% for MRP1-positive patients (P = 0.8;Table 3). Tumor stage (P = 0.001), serum lactate dehydrogenase (P = 0.01) and the International Prognostic Index (0.0001) were also significantly associated with complete response (Table 3). Because LRP expression correlated with tumor stage, we performed another analysis that excluded patients with stage I disease. In this analysis, the complete response rate was 56%for patients without LRP expression and 20% for patients with LRP expression (P = 0.05; data not shown).

Next we performed a logistic regression analysis that included LRP and the International Prognostic Index. In the univariate analysis, the odds ratios for no complete response were 8.3 for LRP(P = 0.006) and 9.5 for the International Prognostic Index (P = 0.0001; Table 4). In the multivariate analysis, the odds ratios for no complete response were 2.3 for LRP(P = 0.4) and 7.6 for the International Prognostic Index (P = 0.004; Table 4).

Overall survival was estimated according to Kaplan-Meier. Fifteen patients died (7 LRP-negative patients, 8 LRP-positive patients). Overall survival was significantly shorter in patients with LRP expression (Fig. 1). At a median follow-up of 2.1 years, median overall survival of all of the patients was not reached. Median overall survival was 0.9 years for LRP-positive patients and was not reached for LRP-negative patients(P = 0.001). With regard to MRP1, 8 MRP1-negative patients and 7 MRP1-positive patients died. Median overall survival was not different between patients with MRP1 expression and those without MRP1 expression (P = 0.9; Fig. 2). In patients with stage II-IV disease(n = 37), overall survival remained significantly shorter in LRP-positive patients than in LRP-negative patients (median 0.9 years versus median not reached; P =0.03; data not shown).

In the univariate Cox regression analysis, the relative risk for death was 4.9 for LRP (P = 0.001) and 4.6 for the International Prognostic Index (P = 0.0001; Table 5). In the multivariate Cox regression analysis that included LRP and the International Prognostic Index, the relative risk for death was 1.4 for LRP (P = 0.6) and 4.0 for the International Prognostic Index (P = 0.005;Table 5).

In the present study, we have determined the expression of LRP and MRP1 in DLBCL and compared the expression with clinical and laboratory parameters of the patients. LRP was positive in 23% and MRP1 in 44%of newly diagnosed DLBCL. LRP expression was associated with poor response to chemotherapy and shorter overall survival, which suggests that LRP is a clinically relevant drug-resistance factor in DLBCL. A similar predictive and prognostic value of LRP expression has previously been reported for acute myeloid leukemia (33, 34, 35, 36), acute lymphoblastic leukemia (37),multiple myeloma (38, 39), and advanced ovarian cancer (40).

LRP strongly correlated with the International Prognostic Index (Table 1). This could explain why LRP lost its statistical significance in the multivariate model. The lack of independent prognostic significance,however, does not mean that LRP is not causative for clinical drug resistance in DLBCL. It also has to be stressed that LRP, but not the factors of the International Prognostic Index, could be altered by therapeutic interventions in the future, thereby improving clinical outcome.

MRP1 expression had no impact on the outcome of chemotherapy or survival of the patients. Similar results have previously been reported in refractory lymphoma patients in which MRP expression has been determined by means of a quantitative PCR assay both before and after chemotherapy (12). MRP1 levels were not different between the pre- and postchemotherapy groups, which suggests that MRP1 overexpression is not responsible for non-P-glycoprotein-mediated drug resistance in these patients. Previous studies in acute myeloid leukemia (35, 36, 41), acute lymphoblastic leukemia (37), and advanced ovarian cancer (40) also failed to demonstrate a predictive or prognostic significance of MRP1 expression.

MDR1/P-glycoprotein expression of lymphomas has been examined in previous studies. Immunohistochemical studies reported P-glycoprotein expression that ranged from 0 to 49% of samples from untreated patients (9, 42, 43, 44, 45, 46, 47, 48, 49). Conflicting results with regard to the clinical importance of MDR1/P-glycoprotein in lymphomas have been reported. MDR1/P-glycoprotein predicted for poor response to induction chemotherapy in two studies (44, 47) but not in two other studies (45, 46). In pretreated lymphomas,MDR1/P-glycoprotein expression was increased because of induction or selection of P-glycoprotein expressing clones (9, 48).

Mutations or overexpression of the p53 gene have been described as predictors of poor response to chemotherapy and shorter survival of lymphoma patients (13, 14, 15). In aggressive B-cell lymphomas, patients with p53 mutations had a lower complete response rate and a shorter overall survival as compared with patients with wild-type p53 (13). A multivariate analysis that included p53 and factors of the International Prognostic Index demonstrated that mutant p53was an independent predictive and prognostic factor (13).

Overexpression of bcl-2 confers drug resistance in vitro by inhibiting apoptosis (50). Although an association between bcl-2 and response to chemotherapy could not be demonstrated (14, 51, 52, 53, 54), bcl-2 expression correlated with a higher relapse-rate (52), shorter disease-free survival (51, 53, 54), and shorter overall survival (53).

In conclusion, LRP expression is associated with poor response to chemotherapy and with shorter survival of the patients with DLBCL and,therefore, may be an important mechanism of drug resistance in this disease. Thus, the development of strategies to clinically overcome LRP-mediated drug resistance should be attempted and might improve clinical outcome in DLBCL in the future.