ALBERT

Description: Background: Acute renal failure (ARF) is a frequent complication of multiple myeloma (MM) and most frequently due to clonotypic light chains (LC) causing cast nephropathy, which is associated with fast deterioration of renal function, increased risk for infections and shortened survival. Here we present the final results of a phase II study employing lenalidomide-dexamethasone as treatment for patients with acute light-chain induced ARF. Patients and methods: 35 patients with LC-induced ARF have been enrolled. Cast nephropathy was confirmed in all 15 patients who had a renal biopsy. Patients with previously unknown MM must have presented with eGFR 〈 50ml/min and serum creatinine ³2.0mg/dL, and those with previously established diagnosis must have had documented eGFR ³ 60ml/min and serum creatinine ≤1.2mg/dL within 6 weeks before deterioration of eGFR to 〈 50ml/min and of serum creatinine to ≥ 2mg/dL due to LC-induced kidney injury. Nine cycles of Lenalidomide, day 1-21, q28 days, with dose adaptation according to eGFR (eGFR 30 – 50ml/min: 10 mg daily, eGFR 〈 30ml/min without requiring dialysis: 15mg q 48 hrs., eGFR 〈 30ml/min requiring dialysis: 5 mg daily following each dialysis) and dexamethasone (Dex), 40 mg, day 1-4, 9-12 and 17-21 during the first cycle and thereafter 40 mg once weekly were planned. Renal response was defined as previously described (Dimopoulos et al, Clin Lymphoma Myeloma. 2009, Ludwig et al. JCO 2010). Results: Patient's median age was: 66 (45-87), 28 patients had newly diagnosed and 7 previously established MM. 5.7% had ISS stage II, 94.3% stage III. 18 patients had light chain myeloma, 14 IgG, and 3 IgA isotype. Adverse cytogenetics t (4; 14) ± del17q ± 1q21 were detected in 14/29 patients. 4/35 patients died and 5 discontinued therapy (3 due to AEs, 1 due to PD, and 1 due to withdrawal of consent) within the first 2 cycles, leaving 26 patients for per protocol (PP) analysis. Median follow up was 17.7 months. Responses were seen in 25/35 (71.4%) patients; 7 (20%) had CR, 3 (8.6%) VGPR, 14 (40%) PR, and 1 (2.9%) MR. Median time to first and to best myeloma response was 28, and 92 days, respectively. Median baseline concentration of involved FLC was 5.465mg/L (range: 147–42.700mg/L) and 8350mg/L (range: 234– 35.500mg/L) in patients reaching ≥PR and ≤MR, respectively, and decreased significantly to a median of 95.75mg/L (range: 11.3–5.630mg/L, p

Description: The prognosis of patients with B-CLL is largely determined by the karyotype of the malignant clone. Microarray technology has facilitated linkage between chromosomal aberrations and gene expression signatures. We have investigated the gene expression profile associated with trisomy 12 (+12). Expression data were obtained by microarray analysis of mRNA from unselected PBMNC of 4 patients with +12 and compared with 16 B-CLL controls. 146 genes were at least 2-fold over- or underexpressed in samples with +12. Five of the 16 genes showing the strongest correlation with +12 were selected for further analysis (HIP1R FC=3,43; MYF6 FC=3,92; P2RY14 FC=−9,59; RASGRP3 FC=−3,85; SLC2A6 FC=2,13) and validation by real time PCR: HIP1R located on chromosome 12q24, with a fold change (FC) of 3,43, MYF6 (chromosome 12q21, FC=3,92), P2RY14 (chromosome 3q21–q25, FC-9,59) RASGRP3 (chromosome 2p25.1–p24.1, FC=−3,85). SLC2A6 (chromosome 9q34, FC=2,13). Quantitative PCR was performed with mRNA from 61 patients (29 with +12, 32 B-CLL controls) and 2 healthy donors. Only 3 genes were significantly associated with +12 compared to the B-CLL-controls in this evaluation: HIP1R (3,486; p

Description: Abstract 2879 Background: Immunomodulatory drugs (IMiDs) are highly active in the treatment of multiple myeloma (MM), but the mechanisms of action are still not completely understood. Recently, cereblon (CRBN) has been identified as the primary target of thalidomide teratogenicity (Ito K et al, 2010) and, moreover as an essential requirement for IMiD therapy (Zhu YX et al, 2011). We wanted to investigate, if expression levels of CRBN could serve as a predictor of response. Patients and Methods: We measured CRBN mRNA expression in bone marrow samples of 44 well characterized MM patients treated with lenalidomide containing regimens, myeloma cell lines, and normal bone marrow (BM), using real time PCR. The median age of patients was 65 years (range: 37–85 years). Nine patients had ISS-stage I, 9 stage II, and 26 had stage III. All patients, except 12, were newly diagnosed. None of the patients had been exposed to lenalidomide before study entry. Full data documentation for response evaluation (〉 2 cycles) was available in 37 patients (84%). Of these, lenalidomide was given in combination with dexamethasone in 27 patients with a starting dose of 25 mg per day on days 1–21 in a 28 days cycle, in combination with melphalan and prednisone (MPR, starting dose of lenalidomide 10 mg per day on days 1–21) in 9 patients, and in combination with bendamustine in 1 patient. Results: Normal BM was used as a reference with an expression level of one. All multiple myeloma cell lines tested (U266, KMS-12-BM, OPM-2, NCL-H929, MM.1S, SK-MM-1, and RPMI8226), had a higher CRBN expression than normal BM. CRBN was detected in all 44 MM samples distributed over a range covering 3 orders of magnitude (0.31 to 462.08-fold relative to normal BM; median: 3.61). Lenalidomide-based therapy resulted in CR in 3 (8%), nCR in 2 (5%), VGPR in 4 (11%), PR in 17 (46%), and in MR in 4 patients (11%), respectively. Three patients (8%) had SD, and 4 (11%) had PD. Median CRBN expression was three times higher in responding (≥MR) patients compared to non-responders (3.65 vs. 0.99, p

Description: Abstract 3947 Introduction: Insulin like growth factor binding protein 7 (IGFBP7) has been described as a secreted tumor suppressor and inducer of apoptotic and senescence pathways, with downregulation in lung cancer, hepatocellular carcinoma, pancreatic carcinoma and other solid tumors, linked to poor prognosis. Recently, a more complex picture has emerged, with IGFBP7 shown to regulate leukemia-stromal cell interactions and to contribute to chemotherapy resistance and leukemia cell survival. IGFBP7 possesses high affinity binding sites for activin A, VEGF-A and insulin and – with lower affinity – to insulin-like growth factor 1 (IGF-1). IGFBP7 has so far not been investigated in multiple myeloma (MM). Our initial gene-expression studies focussing on microenvironmental factors revealed a significant downregulation of IGFBP7 in the MM microenvironment. We therefore aimed to further characterize the role of IGFBP7 in the pathophysiology of MM. Methods: IGFBP7 expression was analyzed by gene expression profiling and quantitative PCR. MMCLs were treated with 5-aza-2x-deoxycytidine, Trichostatin A and recombinant human IGFBP7 for 96 hours. Viability was assessed by the proliferation kit 8. The percentage of apoptotic cells was analyzed by Annexin V/7-AAD staining. Pyrosequencing was performed by Varionostic® GmbH. Immortalized human bone-marrow stromal cells were co-cultured with MMCLs for 72 h. Primary human BMSCs were kept in osteogenic differentiation medium for 7–14 days. Alkaline Phosphatase activity was determined using the AttoPhos® AP Fluorescent Substrate System. Results: Microarray analysis in a large set of MGUS (n = 22) and MM patient samples (n = 329) as well as MM cell lines (MMCLs) (n = 17) demonstrated a significant downregulation of IGFBP7 in each entity compared to normal plasma cells (n = 10). Analyzing the mechanism of IGFBP7 silencing in MM revealed a median upregulation of 2.8 fold after treatment with 5-aza-2x-deoxycytidine and Trichostatin A in 5 of 7 MMCLs (range: 1.8 – 30.2, P 〈 0.05), suggesting that IGFBP7 is controlled via methylation. This was confirmed by pyrosequencing of the IGFBP7 promoter region in 3 primary MM cell samples and 2 MMCLs. We subsequently studied possible functional consequences of IGFBP7 alterations and found that treatment with recIGFBP7 for 96 hours significantly decreased viable cell numbers in 7 of 7 MMCLs tested (relative viability compared to control: 0.68 – 0.92; P 〈 0.05). This effect was due to an impairment of proliferation, as no increase in apoptosis could be detected. Initial data suggest upregulation of cell cycle regulator p21 by IGFBP7 as possibly underlying this effect. Studying the role of IGFBP7 in the MM microenvironment we observed a significant downregulation of IGFBP7 in BMSCs after co-culture with 4 of 5 MMCLs (relative expression compared to control: 0.06 – 0.50; P 〈 0.05). We also studied osteoblast development in vitro and found that IGFBP7 expression is significantly increased during osteogenesis. Treatment with recIGFBP7 further stimulated osteoblast (OB) activity up to 1.8 fold at day 14 (P 〈 0.05). Complementary to these in vitro data, IGFBP7 expression level was associated with the presence of myeloma bone disease in an independent set of bone-marrow CD138+ sorted cells from myeloma patients (n = 62) (P 〈 0.05; see Fig.). Conclusion: Taken together, these results demonstrate that IGFBP7 is downregulated in MM cells by methylation, which likely contributes to loss of cell cycle control and proliferation of malignant plasma cells. IGFBP7 expression seems also to be suppressed in stromal cells in the vicinity of MM cells, which might be involved in the impairment of osteoblast development and contribute to myeloma bone disease. Upregulation of IGFBP7 might be a useful therapeutic intervention in the treatment of MM. Disclosures: No relevant conflicts of interest to declare.

Description: Fludarabine and cyclophosphamide are the backbone of therapy for patients with CLL. The addition of rituximab leads to further improvement of response rates and progression free survival (results from the randomized CLL8 study of the GCLLSG). However, a considerable number of patients have insufficient or short responses to FC or RFC and there is a need to identify factors influencing response or resistance to therapy. The aims of this study are to identify gene expression associated with response or resistance before the start of therapy, to investigate changes in the expression of specific genes or pathways associated with response or resistance during the first cycle of FC and RFC, to provide a rationale for the additional use of novel drugs to improve remission and overcome resistance. We investigated peripheral blood samples from 20 patients receiving FC (n=10) or RFC (n=10) by gene expression profiling, flow cytometry, RT-PCR and western blotting before and during therapy. Sixteen patients received FC or RFC as first line (8 within the CLL8 study) and 4 as second line treatment. All patients were in stage Binet B or C. Gene expression was analyzed and correlated to good (CR or PR) or poor clinical response (SD or PD) at the end of therapy based on NCI-WG/IWCLL criteria. CD19+ cells were harvested by cell sorting before therapy, 24 hours after FC (FC arm), 24 hours after rituximab, and 24 hours after FC (RFC arm). Microarray analysis was performed using Affymetrix U133A gene chips. Genes with a consistent pattern of expression (high or low) in the majority of samples in the good or poor response group were further analyzed. Overall, 9 patients responded adequately to therapy (3 CR, 7 PR), while 11 did not (7 SD, 3PD). Unmutated IgVH status and poor risk cytogenetics were more frequent in poor responders. Gene expression signature before treatment showed that overexpression of 39 genes strongly correlated with response, while overexpression of 20 genes (including HSPA1B, IFI6, APP, CEACAM1, CD9, GAB1, INPP5F) was associated with resistance. Changes in expression after initiation of treatment was also analyzed. Seven genes (including CENTD1, HBA2, COL9A2 and APRIN) were significantly upregulated after rituximab in non-responders. Upregulation of 13 genes (including PMAIP1, SFRS11, CLK1, EFHC1, MRPL39, TUG1, TBRG1, CD49d, PTPRC) after R-FC and 7 genes (including ITPKB, LOC641298, CD44, TAF5) after FC was associated with poor response (resistance) to RFC and FC respectively. Many of these genes are involved in regulation of apoptosis, cell cycle, integrin and PI3-K signaling Therapeutic antibodies or inhibitors against some of these targets are already available. RT-PCR analysis demonstrated a significant downregulation of Akt1 mRNA 24 hours after rituximab infusion in RFC group but no significant changes were observed in patients receiving FC alone. In vitro exposure to rituximab confirmed its in vivo effect and resulted in a significant downregulation of Akt1 and PI3-K-p85 mRNA expression. FACS analysis demonstrated a decrease in the percentage and mean fluorescence intensity (MFI) of surface CD20 after rituximab infusion. This effect was associated with a significant change in total amount and phosphorylation state of CD20 in the RFC group. There was also a decrease in the MFI of CD44 and CD23 after rituximab in the majority of patients in the RFC group but this effect was not consistent in the FC group. In conclusion, we have identified a set of markers associated with good or poor response to FC or RFC before therapy and during the first cycle of treatment. The data provide a rationale for targeted drug combinations to overcome resistance and improve response to therapy in CLL.

Description: We investigated the pattern of lipoprotein lipase (LPL) expression in B-cell chronic lymphocytic leukemia (B-CLL) and assessed its prognostic and potential functional relevance. LPL mRNA levels in peripheral blood mononuclear cells (PBMNC) from 55 CLL patients differed by 2 orders of magnitude between cases with mutated (N=28) or unmutated (N=27) IGVH (median: 1.62 vs. 100.43 compared to normal PBMNC). LPL expression correlated strongly with IGVH mutational status (R=0.78; P

Description: Abstract 2928 Introduction: Bendamustine, an alkylating drug with purine like activities may exert synergistic activity in combination with bortezomib. Here we analyze the efficacy and tolerance of bortezomib-bendamustine-dexamethasone with particular focus on possible discrepancies between patient and physician assessed neuropathic symptoms. Patients: 45 patients with relapsed/refractory MM have been enrolled. Median age: 64 years (range 40–86), male/female: 17/28, ISS stage I/II/III: 14, 16, and 15 patients, respectively. ECOG status 0/I/II: 25, 22, and 2 patients, respectively. Previous treatment lines: 1–2: 25, 3–4: 16, 〉4: 4 patients, respectively. Full data documentation for response evaluation (≥ 2 cycles) is available for 33 patients. Treatment regimen: Bendamustine 70 mg/m2 day 1+4, Bortezomib 1.3 mg/m2 days 1, 4, 8 and 11, Dexamethasone 20 mg on days 1, 4, 8 and 11, repeated every 4 weeks. For assessment of neuropathic sides effects the FACT-GOG/NTX self assessment instrument was used. Efficacy: After a median follow up of 4.7 months, myeloma response (ORR: CR+VGPR+PR) was noted in 17 (51.5%) of the 33 evaluable patients. 5 (15% of) patients achieved CR, 2 (6%) VGPR, 10 (30%) PR; 10 (30%) MR, and 6 (18%) remained stable, while PD was noted in 6 (18% of) patients. Median time to response was 82 days. Responses were observed in 48.1% of patients previously exposed to bortezomib and in 47.1% pretreated with lenalidomide. Responses were seen in 53.8% of patients with high [ampl.1q21, del17p, t (4; 14)] and in 53.8% with standard risk (del13 or no aberration) cytogenetics. Median PFS is 9, 4 months; median overall survival is not reached at present (figure). Tolerance: The regimen was well tolerated with low incidence of infections and gastrointestinal toxicities. Hematological toxicity remained stable from baseline to cycle 4, with G4 anemia, leucopenia and thrombopenia being recorded in 〈 5%. Patient self reported neuropathic symptoms at baseline were recorded as G1–2 in 57% and as G3–4 in 17%, respectively. This pattern remained stable from cycle 1 to 4. In contrast, physician assessed neurotoxicity G1–2 was documented in only 18% of patients at baseline. During the following 4 cycles physician assessed PNP remained constant in almost all patients with only 3 patients developing G3 PNP. G4 PNP was not reported. Conclusions: The BBD regimen yielded a response rate of 52% and a PFS of 9.6 months in heavily pretreated myeloma patients. Substantial response rates were noted in patients pre-exposed to bortezomib (33.3%) and lenalidomide (36.4%) Patient self assessment of neuropathic symptoms revealed a much higher incidence of G1–2 and G3–4 symptoms than physician assessment. Physician assessment of neurotoxicity may underestimate neurologic symptoms associated with disease or neurotoxic treatment. Disclosures: Ludwig: Mundipharma, Janssen-Cilag: Research Funding, Speakers Bureau. Greil:AOP Orphan Pharmaceuticals AG: Research Funding.

Description: Background and Aims: Monokine-induced by interferon-γ (MIG) is a chemokine that is produced by monocytes and macrophages in response to interferon-γ and acts as a chemoattractant mainly to T-lymphocytes in inflammatory processes. MIG has also been suggested to act in an autocrine loop to stimulate tumour cells through its receptor CXCR3, which is known to be expressed in myeloma cells. However, it is presently unclear if MIG is of biologic significance in myeloma in vivo. We have recently shown that multiple myeloma oncogene 1 (MUM1) expression in myeloma cells correlates with prognosis in this disease (Heintel D et al., ASH 2005), and MUM1 was found to upregulate MIG gene expression in B cell malignancies (Uranishi M et al., Leukemia 2005). This led us to evaluate the potential prognostic significance of MIG serum levels in a series of well characterized myeloma patients. Methods: 105 newly diagnosed multiple myeloma patients (median age 69.3 years, range 39.4–90.5) were enrolled. 18 patients presented with Durie/Salmon stage I disease, 9 had stage II and 78 had stage III. MIG serum levels were determined by a commercially available ELISA (R&D Systems). Serum samples from 17 MGUS patients and 37 age-matched healthy volunteers were used as controls. Results: MIG serum levels were elevated in multiple myeloma patients (median 161.3 pg/ml, range 9.37–1966.0) compared to MGUS patients (median 92.7 pg/ml, range 6.29–1303.1) and healthy controls (median 106.2 pg/ml, range 51.0–390.6). For analysis of myeloma cases, a cut-off level for MIG of 200pg/ml (=95th percentile of MIG in controls) was chosen to identify low and high MIG expressers. Using the cut-off as defined above, 63 patients with low MIG serum levels and 42 high MIG expressers were identified in a population of 105 myeloma patients. MIG serum levels in myeloma patients showed strong correlations with several markers of tumour load including low albumin and high β2-microglobulin. Interestingly, no correlation was found with C-reactive protein levels, indicating that MIG is not associated with an inflammatory response in myeloma. Median survival was significantly shorter in patients with high MIG serum levels compared to patients with low MIG expression (median not reached vs. 17.0 months, p

Description: Abstract 4031 Introduction: Excessive production of free light chains with affinity for uromodulin results in protein aggregates and toxic injury of distal renal tubules. The ensuing renal failure is a significant risk factor for infections, dependency on chronic hemodialysis and reduced survival. Management of this emergency includes rapid confirmation of the diagnosis and prompt installment of effective anti-myeloma therapy. Here, we assess the efficacy of lenalidomide-dexamethasone for treatment of patients with LC-ARF. Patients and Methods: 32 patients with LC-ARF as formerly defined (J. Clin. Oncol. 2010 20; 28(30):4635-41) have been enrolled so far. Age (median): 66 years (range: 46–87 years), Gender: male/female: 17/15. All patients presented with ISS stage III. 26 (81.3%) had de novo MM and 6 (18.8%) previously treated, but relapsing disease. Median GFR was 19.9 ml/min (range 6.1 – 37.2 ml/min). ECOG performance status was 0 in 9, I-II in 18 and III-IV in 5 patients, respectively. Lenalidomide was given from d 1–21 with dose adaptation according to GFR as suggested in the prescribing information. Dexamethasone 40 mg was administered on d 1–4, 9–12, 17–20 during cycle 1; thereafter 1x/week. Cycles were repeated q 4 weeks. Results: Presently, 23 patients are evaluable for response (completed ≥2 cycles and fully documented). The median follow-up is 7.7 months, median number of cycles is 9 (range 2–9). CR was achieved in 5 (21.7%), nCR in 1 (4%), VGPR in 2 (8%) and PR in 13 (52%) patients, MR in 1 (3%), respectively, yielding an ORR (CR+nCR+VGPR+PR) of 91.3% for evaluable patients and 65.6% for the ITT population. Median time to first myeloma response was 28 (range 27–63 days) and to best response was 113 days (34–304 days). The cumulative incidence of all patients with myeloma and renal responses are shown in figure 1. Median PFS and OS were 13.8 and 31.2 months respectively in the evaluable patients and 7.4 and 31.2 months in the ITT population. Renal response was assessed as formerly defined (J Clin Oncol. 2010 20; 28(30):4635-41). 4 patients achieved CRrenal, 8 PRrenal and 3 MRrenal, yielding an ORRrenal in 15 patients (65.2% of the evaluable and 46.9% of the ITT population). Median time to first renal response was 28 (range: 27–34) days, and to best renal response 119 days (34–304 days). 5 of 13 dialysis dependent patients became dialysis independent. Median GFR of evaluable patients increased from 15.2 (range 6.1 – 35.1 ml/min) at baseline to a median best GFR of 31.4 ml/min (range 11.3 – 103.2 ml/min). In the 5 patients with CR a significant increase in GFR (median 26.7 to 60.9 ml/min) and in the 16 patients with nCR/VGPR/PR an increase from 13.5 to 30.1 ml/min was observed. Full documentation of adverse events is presently available in 32 patients. 5 patients died within the first 2 months, 2 (8.7%) each due to infection and cardiac arrest and 1 (4.3%) with apoplexia. Grade 3/4 anemia, thrombopenia and leucopenia, were seen in 17 (53.1%), 9 (28.1%), and 5 (15.6%) patients, respectively. Other common grade 3/4 toxicities were infection/sepsis in 13 (40.6%), and cardiac dysfunction in 8 (25%) patients, respectively. Exanthema G3 was seen in 3 patients (9.3%), pulmonary embolism, macula edema and multiple stroke syndrome in 1 (3.1%), potassium deficiency G3/4 in 5 (15.6%), and oral candidiasis in 2 patients (6.3%) each. Conclusions: LD showed significant anti-myeloma activity with an overall myeloma response rate (CR-PR) of 91.3% in the evaluable of 65.5% in the ITT cohort. Renal responses (CRrenal-PRrenal) were observed in 65.2% and 46.9% patients, respectively. Time to first myeloma and first renal response was fast (28 days each). The LD regimen with the lenalidomide dose adjusted to GFR was well tolerated. Updated results will be presented. Disclosures: Ludwig: Janssen Cilag: Honoraria; Mundipharma: Honoraria; Celgene: Honoraria. Off Label Use: Lenalidomide was used of label in combination with dexamethasone in this phase II study in patients with acute light-chain induced renal failure.

Description: MUM1 (multiple myeloma oncogene 1)/IRF4 (interferon regulatory factor 4) has been shown to be expressed by neoplastic B-cells and to be closely linked to poor survival in large B-cell lymphomas. Information on a possible expression of MUM1 in myeloma cells and on a possible role as prognostic marker has not been presented up to now. Hence, we aimed to analyse these issues in a cohort of well characterized patients with multiple myeloma. Fifty-three patients with multiple myeloma, median age 70 years, were enrolled. Nine presented with Durie Salmon stage I, 6 had stage II, and 38 stage III. Forty-five were newly diagnosed and 8 pretreated (median number of treatment lines: 2). MUM1 mRNA expression was analysed by real time PCR in freshly isolated unsorted as well as CD 138 enriched myeloma cells and controls. Survival was calculated using the Kaplan-Meier product limit method. For multivariate analyses the Cox proportional hazards method was used. MUM1 mRNA levels were detected in all 53 unsorted myeloma samples with a distribution range covering 3 orders of magnitude (0.35 to 261.38-fold, median: 10.5 relative to normal blood=1). MUM1 RNA expression was validated in CD138 enriched myeloma cells (N=10). A strong but heterogeneous expression was found in the CD 138 positive cell fraction while low expression was found in the remaining CD138 negative cell fraction. When patients were divided regarding the magnitude of MUM1 expression by using a cut off level of 13 (relative to the level detected in normal blood) 32 were found with low and 21 with high MUM1 expression. Median survival was significantly shorter in patients with high MUM1 expression compared to low expressers (45 months, vs. 58 months, p =8x10−5, Hazard Ratio 9.65). Among the following parameters: ß2-microglobulin (〉5mg/ml), CRP, albumin, creatinine, M-component, and bone marrow plasma cell infiltration tested, only increased MUM1 expression (p = 0.00088), high ß2-microglobulin (〉5mg/ml, p = 9x10−4), creatinine (p 5mg/ml and MUM1 expression 〉13) had very short survival (median: 3 months). In contrast median survival was 58 months in the 41 patients with none or only one adverse prognostic factor (p = 6.71 x 10−11). In conclusion, high MUM1 RNA expression in myeloma cells was found in 40% of patients. High MUM1 RNA expression strongly correlated with poor survival both in univariate and in multivariate analysis. The latter analysis revealed MUM1 RNA expression together with high ß2-microglobulin levels as only parameters independently predicting for poor outcome.