|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 22, No 9 (May 1), 2004: pp. 1664-1673 © 2004 American Society of Clinical Oncology. DOI: 10.1200/JCO.2004.06.105 Influence of Biologic Markers on the Outcome of Hodgkin's Lymphoma: A Study by the Spanish Hodgkin’s Lymphoma Study GroupFrom the Medicina Interna, Hematología, and Unidad de Bioestadística Clínica, Hospital Ramón y Cajal; Lymphoma Group, Molecular Pathology Program, Centro Nacional de Investigaciones Oncológicas; Hospital Universitario 12 de Octubre; Hospital Clínico Universitario San Carlos; Hospital General Universitario Gregorio Marañón; Hospital de Móstoles, Madrid; Hospital Clínico Universitario, Santiago de Compostela; Hospital Marqués de Valdecilla, Santander; Hospital Virgen de la Salud, Toledo; Hospital Clínico Universitario, Salamanca; Hospital Universitario de Canarias, Tenerife; Complejo Hospitalario Xeral-Cies, Vigo; Hospital Central de Asturias, Oviedo; and Hospital Severo Ochoa, Leganés, Spain. Address reprint requests to Juan F. García, MD, PhD, Molecular Pathology Program, Centro Nacional de Investigaciones Oncológicas (CNIO), C/O Melchor Fernández Almagro, 3 E-28029 Madrid, Spain; e-mai: jfgarcia{at}cnio.es
PURPOSE: Current therapies fail to cure a significant proportion of patients with Hodgkin's lymphoma (HL). Predictive systems for stratification of the disease and selection of treatment based on sets of clinical variables, such as the international prognostic score (IPS), are of relatively small practical value. The predictive use of biologic parameters has so far provided limited and inconsistent results. Here we explore the influence of a set of molecular markers on the outcome of HL. PATIENTS AND METHODS: Forty molecular markers involved in B-cell differentiation and activation, signal transduction, cell cycle, and apoptosis control were analyzed in 259 classic HL patient cases by using tissue microarrays. Univariate analysis was performed to evaluate the influence of markers on favorable outcome (complete remission of > 12 months). Significant variables were included in a multivariate logistic regression analysis, and the probability of favorable outcome was estimated. RESULTS: Univariate analysis revealed four molecular markers that predicted outcome, and the multivariate analysis showed p53, Bcl-XL, and terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate-biotin nick-end labeling (TUNEL) to have independent significance. The combination of these factors determined two groups of patients (group I, zero to one factor; group II, two to three factors) with a probability of a favorable outcome of .948 and .687, respectively. A multivariate Cox's model shows that these biologic risk groups have special predictive power in low-IPS patients. CONCLUSION: The data from this exploratory study suggest that the accumulation of molecular events seems to influence the outcome of HL, particularly in the low-IPS group.
Even though Hodgkin's lymphoma (HL) is a curable tumor in most cases, approximately 20% to 30% of all patients experience relapse and eventually die as a result of progressive disease or complications of therapy.1-3 In contrast, another fraction of patients may be overtreated with unnecessarily aggressive chemotherapy protocols.2 Therefore, an accurate prediction of the results of treatment might eventually allow the identification of patients who are likely to benefit from reduced therapy, or those who have a low probability of having a sustained response to standard treatment. The identification of clinical and analytic prognostic factors in HL has been studied thoroughly.2,4-6 However, most prognostic systems used to date, including the international prognostic score (IPS),2 fail to identify with sufficient accuracy the proportions of patients with favorable or unfavorable responses to treatment. Although the IPS is suitable demonstrably for most HL patients with advanced disease,7,8 its effectiveness for prediction at earlier stages is more controversial. Thus, clinical outcome is assessed erroneously in almost one third of HL patients when predictors based solely on clinical data are used.6,7 Previous studies of different types of lymphoid and epithelial neoplasms illustrate how the accumulation of genetic and epigenetic alterations is associated with increased clinical aggressiveness and limited responses to standard chemotherapy.9 A number of studies using high-throughput technologies have revealed the power of biologic scoring for predicting outcome and therapy response in several tumoral models.10-12 The alterations in the main pathways responsible for controlling the cell cycle and apoptosis machinery in HL are still poorly understood.13 However, studies have shown some relationships between the expression of activation or differentiation markers in the neoplastic Hodgkin's and Reed-Sternberg (H/RS) cells,14,15 cell cycle and apoptosis deregulation,13,16,17 signal transduction pathways activation,18 host response,19-21 Epstein-Barr virus (EBV) detection,15,22-24 and other biologic factors with the clinical outcome of HL patients. Confirming the existence of the large number of molecules putatively associated with clinical outcome in HL samples and analyzing these relationships require high-throughput techniques. Tissue microarrays (TMAs) allow simultaneous analysis of several dozen proteins in large series of patients, thus revealing interactions between diverse pathways and genes, and in turn, the clinical relevance of multiple biologic variables.25 The robustness and reproducibility of this technique for analyzing HL have been demonstrated previously.13,26,27 Given the complexity of the biologic processes and the different molecular markers that are claimed to predict the clinical outcome of HL patients, we decided to use TMA technology in a series of 259 patients to evaluate the expression of a set of 40 biologic markers involved in cell differentiation and activation, cell cycle control, apoptosis regulation, and host response. The results obtained were used to derive biologic risk groups (BRGs) to explore the possible influence of the accumulation of molecular markers in the response to treatment and outcome of HL patients.
Patients Clinical data and biopsy samples from 447 patients were retrospectively collected from the participating member centers of the Spanish Hodgkin's Lymphoma Study Group. Some of the patients in this series have been included in a previous report.13 All of the clinical and follow-up data were reviewed by two participating clinical members (C.M., L.G.C.) and apparent errors and missing values were reported to the contributors for confirmation or correction. Data recorded in the database included sex, age, Ann Arbor stage, presence of B symptoms, bone marrow status, and analytic variables included in the IPS2 (hemoglobin and albumin levels, and leukocyte and lymphocyte counts). Patients that met the following criteria were included: initial diagnosis of classical HL performed in a lymph node biopsy before any form of treatment; an available and suitable diagnostic paraffin-embedded tissue block for histologic revision and TMA analysis; HIV-negative status (this condition was determined either by HIV serology using enzyme-linked immunosorbent assay or immunohistochemical staining for p24 protein28); any Ann Arbor stage; any age; and management with standard treatment modalities adapted to clinical stages. From the original 447 patient cases collected, 259 patient cases fulfilling these criteria remained after clinical and diagnostic data validation, and were included in the statistical analyses. The rejected patients were excluded because they lacked essential clinical or follow-up data, or more commonly, because the quality or amount of the remaining tissue in the paraffin block prevented its use for the TMA analysis. Treatment decision depended on the criteria of the local physician, according to the current established protocols for the treatment of HL (Table 1). Patients with advanced HL were treated mainly with six to 10 courses of combination chemotherapy including doxorubicin (doxorubicin, bleomycin, vinblastine, dacarbazine or variants; 174 patients); only six of 180 patients received chemotherapy that did not include doxorubicin (mechlorethamine, vincristine, procarbazine, prednisone, or variants). Patients with initial bulky masses or with residual masses with any suspicion of activity received involved-field radiotherapy. Low-risk patients (stages I to IIA) received extended-field radiotherapy or two to four courses of chemotherapy and involved-field radiotherapy. A total of 36 patients (13.8%) were treated solely with radiotherapy, 217 patients (83.8%) were treated with doxorubicin-containing chemotherapy (113 of whom also received some form of radiotherapy), and six patients (2.3%) were treated with mechlorethamine, vincristine, procarbazine, prednisone or variants (one of these patients also was treated with radiotherapy). Nine advanced-stage patients received autologous peripheral stem-cell transplantation either as consolidation therapy or as escalating treatment after partial response.
TMA Design Immunohistochemical (IHC) expression and in situ hybridization (ISH) of the different biologic markers were assessed using TMA technology. To this end, we used a Tissue Arrayer device (Beecher Instruments, Sun Prairie, WI), as previously described.13,25 Several studies have shown this procedure to be suitable and reproducible for analyzing HL series. Specifically, several markers included in the multivariate analysis have yielded results highly concordant with those of whole-section analyses.13,26,27 Additional details for the selection of core samples and evaluation of the results have been described.13
IHC and ISH Analyses
Apoptosis was detected using the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Intergen Co, Oxford, United Kingdom), as described elsewhere.13 EBV was detected by ISH with fluorescein-conjugated EBV PNA probe (DAKO), following the manufacturer's recommendations. Stained TMA sections were evaluated by two different pathologists (J.F.G. and M.A.P.) using uniform criteria. To guarantee the reproducibility of this method we used the same criteria as previously described.13 The staining pattern for each antibody was recorded as positive or negative, and high or low level of expression, taking into account the expression in H/RS cells or in reactive benign cells, and using the various previously established cutoffs.13
Statistical Analysis For the study of survival, the end point was disease-specific survival (DSS), considering that deaths resulting from the complication of treatment did not reflect the impact of the molecular factors in the outcome of the disease. DSS was taken as the interval from the beginning of treatment to death as a result of any cause directly related to the disease. Patients known to have died from causes unrelated to HL or its treatment were therefore censored from the survival analysis. Disease-free survival (DFS) refers to the lapse from remission to relapse or death as a result of causes directly related to HL. DFS was used only to select the groups with favorable or unfavorable outcome in patients achieving CR. Event-free survival (EFS) was defined as the interval between initial therapy and first recurrence; patients who had not achieved CR were therefore assigned an EFS of zero, following a commonly used criteria.29,30 To evaluate the final impact of the clinical and biologic variables in this population of patients, two groups of patients were considered: a group with favorable outcome (favorable response [FR]), consisting of the patients who achieved CR and maintained it for 12 months, and patients with an unfavorable outcome (unfavorable response [UR]), consisting of those who either did not achieve CR or experienced relapse less than 12 months after CR. This end point was selected because in advanced HL patients, it has been demonstrated repeatedly to be a close surrogate of the final outcome of the disease.1,31-34 We also adopted this end point for early HL because the molecular variables might have a critical influence on the final outcome, independently of the staging or clinical status at diagnosis.13,35
For univariate analysis, the A multivariate logistic regression model using the molecular variables36 was used as a first step toward establishing the risk groups. In keeping with the proposal of the study to explore the predictive value of biologic variables, only molecular data were used for the multivariate analysis. Treatment response (FR or UR) was the dependent variable. Only those variables that had been significant in the univariate analysis were included in the maximum multivariate model. The final, most parsimonious model was obtained by backward selection from the maximal model. The log-likelihood ratio test was used for model comparison and goodness of fit assessment. Colinearity was tested using Belsley's criteria.37 The final model was internally validated, including calibration and discrimination, using the leave-one-out method.38 The area under the receiver operating characteristic curve39 and the Hosmer-Lemeshow36 statistic were then calculated. Multivariate Cox models were used to compare the predictive performance for DSS of the biologic groups with the IPS prediction. The proportional hazards assumption was tested by introducing variables into the model created by multiplying each variable by time.40 Multivariate Cox analysis was also used to evaluate EFS.
The influence of the two risk groups on DSS and EFS, overall and stratified by IPS, was estimated by the Kaplan-Meier method and log-rank test in the whole series, in patients with IPS 0 to 2 or All tests were two-sided and P values of .05 or less were considered statistically significant. The SPSS program (SPSS Inc, Chicago, IL; 1999) was used for these analyses.
The main clinical features of the 259 patient cases are listed in Table 1. The median age at diagnosis was 38 years (range, 10 to 86 years). At initial presentation, stage III and IV disease was present in 37% of the patients, 73% of whom had low IPS (0 to 2). Most patients (58.7%) presented with the nodular sclerosis subtype of HL. The median period of follow-up was 49 months.
Univariate Analysis The results of the univariate analysis are listed in Table 3. Overall, there was considerable concordance between molecular variables associated with DSS and with treatment response. Specifically, in the analyses of treatment response (FR or UR), only p53, p21, terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate-biotin nick-end labeling (TUNEL), and Bcl-XL were statistically significant. Significant associations of IPS with EFS, DFS, and FR also were observed. The comparison between the different variables in this univariate analysis was considered worthwhile only in groups with at least 10 patients in each category. Therefore, comparison was not possible for CD30, multiple myeloma-1/interferon regulatory factor-4, cyclin A, and Bax.
Multivariate Logistic Regression Analysis In the multivariate logistic regression model only 239 of 259 patients (209 with FR and 30 UR) had at least 12 months of follow-up and were therefore included in this analysis. The maximal model included TUNEL, p21, Bcl-XL, and p53. The final model features only p53, Bcl-XL, and TUNEL as independent significant factors. Table 4 lists the estimated coefficients, their SEs, the P values corresponding to Wald's test for each coefficient, and the log-likelihood ratio test for the model. The model was internally validated by the leave-one-out method. The area under the receiver operating characteristic curve in the validation model was 0.659, which was slightly smaller than in the original model (0.770). Table 5 lists observed and predicted event rates for different risk levels and the Hosmer-Lemeshow's  2 statistic, revealing the calibration to be excellent.
The model predicts the FR probability of each patient, and from this probability we classified patients into two risk groups (BRGs): patients with predicted probabilities for FR  .90 (group I), who are those with none or only one adverse molecular marker (p53, Bcl-XL, and TUNEL); and patients with predicted probabilities less than .90 (group II), who are those with two or three markers. Table 6 lists the number of patients in each category, the favorable outcome probability predicted by the model, the resulting groups, and the FR probability predicted by a new model, considering only the group variable. Using this final model, the predicted probabilities for FR are .948 (group I) and .687 (group II).
To validate the ability of BRGs to predict DSS, they were applied to the entire series of 259 patients (including those patients not included in the multivariate logistic model): 185 patients (71.4%) were classified in group I, and 74 patients (28.6%) were classified in group II. DSS curves for the two groups are shown in Figure 1A. These groups had different survival during the follow-up period (log-rank, 22.37; P = .0000), and their 5-year survival probabilities (group I, .94; group II, .66;) were close to the FR probability predicted by the logistic model (Table 6).
Figure 1B shows the DSS curves for IPS stratification for the 254 patients with sufficient data to allow classification by the IPS (log-rank, 15.33; P = .0001). Figures 1C and 1D also show the DSS curves of the two groups stratified by IPS, showing that this BRG discriminated different risk groups in the low-IPS patients (log-rank, 25.52; P = .0000; Fig 1C) but not in the high-IPS patients (log-rank, 0.58; P = .45; Fig 1D). Figure 1E shows that the BRG also discriminated two groups in patients treated by chemotherapy regimens containing doxorubicin (log-rank, 15.36; P = .0001). Figure 2 shows the EFS curves, confirming that BRG significantly distinguishes HL patients, when considering all the series (log-rank, 4.54; P = .033; Fig 2A), or exclusively patients treated with doxorubicin-containing regimens (log-rank, 4.58; P = .0324; Fig 2E). The BRG could not predict EFS significantly when patients were stratified by IPS (Figs 2C and 2D). The EFS was significantly different among the IPS groups (log-rank, 13.33; P = .003).
To confirm the interaction between the two classification systems, a Cox multivariate model for DSS was fitted, including the predicted group (BRG), IPS, and their interaction as independent variables. The interaction was significant, confirming that BRG discriminates groups of different prognosis in low-IPS patients, but not in high-IPS patients. In low-IPS patients, the mortality risk for group II was 12.5-fold (95% CI, 3.5 to 44.9) that of group I, whereas for high-IPS patients this hazard ratio was not significant (1.58; 95% CI, 0.6 to 4.1; Table 7).
Multivariate Cox Analysis of EFS To demonstrate the time-dependent predictive capacity of biologic variables, an analysis using the multivariate Cox model was performed. This included a different set of variables significantly associated with EFS from those revealed by the univariate analysis: p53, TUNEL, and T-cell intracellular antigen-1 (nontumoral component). The final model revealed T-cell intracellular antigen-1 (nontumoral) and p53 as independent predictive markers (P = .022 and 0.044, respectively), with a hazard ratio of 2.37 (95% CI, 1.29 to 4.37) for the BRG created from this model. This new model also significantly distinguishes two risk groups when considering all the series for EFS (P = .0043) and DSS (P = .0014) using Kaplan-Meier analysis (data not shown). These results confirm the general predictive power of the biologic markers. However, the statistical independence of the IPS for this end point (EFS) cannot be evaluated technically, given that the IPS stratification did not fulfill the proportional hazards assumption in this series of patients.
HL is an apparently homogeneous tumor with a relatively regular clinical course and a favorable outcome in most patients.41 This apparent uniformity permits a rather clear-cut staging of the disease, defining groups with increasing extension and aggressiveness that require therapeutic approaches with different intensities of treatment.2,42 However, up to one third of the patients did not conform to this uniformity and did not respond favorably to standard treatments. The reasons for this poorer outcome might be assumed to depend on the specific molecular mechanism influencing the therapy response.17,35,43 Clinical factors, such as bulky disease, advanced clinical stages, or B symptoms, are surrogate markers of the biology of the tumor and have been demonstrated to be classic prognostic markers.17,44 This also is the case with acute reactants, such as high erythrocyte sedimentation rate, lactate dehydrogenase, beta2-microglobulin, and others.44,45 Moreover, the combination of easily obtainable clinical and analytic variables included in the IPS2 allows discrimination of groups with different outcomes.
Different research groups have explored the potential use of biologic markers as determinants of clinical outcome, and the prognostic influence of some of them has been well assessed. Cell cycle and apoptosis regulators, signal transduction pathways, and differentiation or activation markers have been postulated as prognostic predictors because they condition the aggressiveness of the tumors: altered p53 protein expression,13,16,17 deregulation of members of the Bcl2 family,13,17,46,47 proliferative and/or apoptotic indexes,13,48,49 activation of the NF- TMA technology allows simultaneous analysis of several proteins in large series of patients. Using this tool, key alterations in cell-regulation factors and apoptotic pathways have been demonstrated previously in HL.13,26,27 Because these molecular alterations are essential to the behavior and outcome of HL, we have explored their prognostic influence in this article. In general, the results obtained from the univariate analyses are highly consistent for most biologic markers, showing overall parallelism between DSS and treatment response for significant variables. Our results show that the expression in H/RS cells of Bcl-XL, p53, and apoptosis determined by the TUNEL technique has independent prognostic significance and that the accumulation of these factors can accurately separate HL patients into two risk groups (BRGs), which significantly influence the outcome. Alternatively, a multivariate Cox model for EFS also could demonstrate the predictive capacity of biologic variables. However, comparison with the IPS prediction was not possible because of the lack of proportionality, and in this study, we therefore selected the multivariate logistic model for FR.
To validate the biologic model, we studied the influence of the biologic groups in the survival of the 259 patients of the series and also when stratified into groups with low IPS (0 to 2) and high IPS ( Interestingly, in the low-IPS group there were two types of patients depending on the molecular profile; that is, patients with apparently good prognosis as indicated by the IPS may have had tumors harboring molecular abnormalities that conferred a poorer outcome. However, these differences were not found in the group with higher IPS. These results are encouraging, but additional studies of the significance of the aggregation of these (and probably other) molecular factors are needed in larger series of patients. The individual unfavorable prognostic influence of each of these three biologic factors has been reported repeatedly. Abnormal p53 expression is a classic adverse prognostic marker in different lymphoid malignancies.50,51 Nuclear overexpression of p53 protein is a common phenomenon in HL, which, despite being associated with the presence of wild-type p53 gene in most cases,52,53 probably reflects disturbances in the p53-Hdm2-p14ARF pathway.54 Several authors have reported that in HL an abnormal p53 expression is associated with shorter overall survival and/or DFS.13,16,17
The presence of multiple alterations in apoptosis regulatory molecules is a frequent finding in cancer;9,35,51 the molecules seem to play a role in treatment resistance. Overexpression of Bcl2 protein,13,15-17,47 and other Bcl2 family members such as Bcl-XL13 and Bax,46 in addition to other apoptosis regulators,55,56 have been described in HL. A direct relationship between Bcl2 overexpression and FFS has been described specifically in HL.47 In this study, we found a tendency toward shorter DSS and EFS in Bcl-2–positive HL patients, although the differences were not significant. However, in this study we have observed the significant predictive power of Bcl-XL protein, consistent with the findings of other studies,16,18 suggesting that Bcl2, Bcl-XL, and NF- Finally, the high apoptotic index measured by TUNEL also has been found to constitute an independent prognostic factor in our series, corroborating previous observations.13,48,49 The independent predictive capacity of this marker is notably higher than that of the other two markers, p53 and Bcl-XL protein overexpression. Our study failed to confirm other previous observations, notably those concerning Rb, Ki67, and EBV expression.24 The discrepancy with previous studies may be related to the different composition of the series and the different techniques used for their analysis, and we cannot rule out the possibility that it is specifically linked to the use of TMAs in this study, given that TMAs are known to have greater sensitivity and to produce more reproducible results.13,26,27 In conclusion, our results suggest that the accumulation of molecular events seems to be associated with a worse outcome; specifically, two different outcomes within the low-IPS group are discriminated. Whether such discrimination is possible in patients with a high IPS and whether other molecular alterations might also have some influence needs to be established in additional studies. This approach using multiple molecular variables should be validated in a larger series to evaluate accurately their influence on the outcome of HL, and to identify groups with a different prognosis that would require individualized forms of therapy.
The following members of the Spanish Hodgkin's Lymphoma Study Group participated in the study: P. Domínguez, C. Jara (FHA, Alcorcón), M.J. Mestre, R. Quibén, M. Méndez, L. Borbolla (H Móstoles, Madrid), M.A. Martínez, C. Grande (H 12 Octubre, Madrid), C. Bellas, C. Montalbán, J. García-Laraña (H Ramón y Cajal, Madrid), A. Castaño, P. Sánchez-Godoy (H Servero Ochoa, Leganés), C. Martín, R. Martínez (HUC San Carlos, Madrid), J. Menárguez, P. Sabín, E. Flores (H Gregorio Marañón, Madrid), J. González-Carrero, C. Ponderós (H Xeral-Cies, Vigo), T. Álvaro, Ll. Font (H Verge de la Cinta, Tortosa), M. Mollejo, M.A. Cruz (H Virgen de la Salud, Toledo), H. Álvarez-Arguelles, M. Llanos (HU Canarias), C. Morante (H Cabueñes, Gijón), F. Mazorra, E. Conde (HM De Valdecilla, Santer), M.F. Fresno, C. Rayón (HC de Asturias, Oviedo), T. Flores, R. García-Sanz (HCU Salamanca), J. Guma (H Sant Joan, Reus), P. Gonzalvo (HC de Jarrio, Coaña), G. Fernández (H Alvarez Buyllas, Mieres), J. Forteza, M. Fraga, J.L. Bello (F Med Santiago de Compostela), J.R. Méndez (H Valle de Nalón, Asturias), J.F. García, M.M. Morente, and M.A. Piris (CNIO, Madrid). Bioinformatics and data analysis: V. Abraira, J. Zamora (H. Ramón y Cajal). Data management: L. Cereceda, L. González-Camacho, I. Cuenca (CNIO Tumor Bank).
The authors indicated no potential conflicts of interest.
We thank Laura Cereceda and Irene Cuenca (CNIO Tumor Bank Unit) for their excellent assistance with data management, and Raquel Pajares, Ana Díaz, and Maria J. Acuña for their technical assistance.
Supported by grants from the Fondo de Investigaciones Sanitarias (FIS PI020323 and G03/179) and the Ministerio de Ciencia y Tecnología (SAF2001-0060), Spain. Carlos Montalbán and Juan F. García contributed equally to this work. Authors' disclosures of potential conflicts of interest are found at the end of this article.
1. Sureda A, Arranz R, Iriondo A, et al: Autologous stem-cell transplantation for Hodgkin's disease: Results and prognostic factors in 494 patients from the Grupo Espanol de Linfomas/Transplante Autologo de Medula Osea Spanish Cooperative Group. J Clin Oncol 19:1395-1404, 2001
2. Hasenclever D, Diehl V: A prognostic score for advanced Hodgkin's disease: International Prognostic Factors Project on Advanced Hodgkin's Disease. N Engl J Med 339:1506-1514, 1998 3. Pfreundschuh MG, Rueffer U, Lathan B, et al: Dexa-BEAM in patients with Hodgkin's disease refractory to multidrug chemotherapy regimens: A trial of the German Hodgkin's Disease Study Group. J Clin Oncol 12:580-586, 1994[Abstract] 4. Ferme C, Bastion Y, Brice P, et al: Prognosis of patients with advanced Hodgkin's disease: Evaluation of four prognostic models using 344 patients included in the Group d'Etudes des Lymphomes de l'Adulte Study. Cancer 80:1124-1133, 1997[CrossRef][Medline] 5. Vassilakopoulos TP, Angelopoulou MK, Siakantaris MP, et al: Prognostic factors in advanced stage Hodgkin's lymphoma: The significance of the number of involved anatomic sites. Eur J Haematol 67:279-288, 2001[CrossRef][Medline] 6. Gobbi PG, Zinzani PL, Broglia C, et al: Comparison of prognostic models in patients with advanced Hodgkin disease: Promising results from integration of the best three systems. Cancer 91:1467-1478, 2001[CrossRef][Medline] 7. Zander T, Wiedenmann S, Wolf J: Prognostic factors in Hodgkin's lymphoma. Ann Oncol 13:67-74, 2002[Abstract]
8. Bierman PJ, Lynch JC, Bociek RG, et al: The International Prognostic Factors Project score for advanced Hodgkin's disease is useful for predicting outcome of autologous hematopoietic stem cell transplantation. Ann Oncol 13:1370-1377, 2002 9. Sanchez E, Chacon I, Plaza MM, et al: Clinical outcome in diffuse large B-cell lymphoma is dependent on the relationship between different cell-cycle regulator proteins. J Clin Oncol 16:1931-1939, 1998[Abstract] 10. Shipp MA, Ross KN, Tamayo P, et al: Diffuse large B-cell lymphoma outcome prediction by gene-expression profiling and supervised machine learning. Nat Med 8:68-74, 2002[CrossRef][Medline]
11. Rosenwald A, Wright G, Chan WC, et al: The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med 346:1937-1947, 2002
12. van de Vijver MJ, He YD, van't Veer LJ, et al: A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 347:1999-2009, 2002
13. Garcia JF, Camacho FI, Morente M, et al: Hodgkin and Reed-Sternberg cells harbor alterations in the major tumor suppressor pathways and cell-cycle checkpoints: Analyses using tissue microarrays. Blood 101:681-689, 2003 14. Benharroch D, Dima E, Levy A, et al: Differential expression of sialyl and non-sialyl-CD15 antigens on Hodgkin-Reed-Sternberg cells: Significance in Hodgkin's disease. Leuk Lymphoma 39:185-194, 2000[Medline] 15. Vassallo J, Metze K, Traina F, et al: The prognostic relevance of apoptosis-related proteins in classical Hodgkin's lymphomas. Leuk Lymphoma 44:483-488, 2003[CrossRef][Medline] 16. Brink AA, Oudejans JJ, van den Brule AJ, et al: Low p53 and high bcl-2 expression in Reed-Sternberg cells predicts poor clinical outcome for Hodgkin's disease: Involvement of apoptosis resistance? Mod Pathol 11:376-383, 1998[Medline]
17. Smolewski P, Robak T, Krykowski E, et al: Prognostic factors in Hodgkin's disease: Multivariate analysis of 327 patients from a single institution. Clin Cancer Res 6:1150-1160, 2000
18. Hinz M, Loser P, Mathas S, et al: Constitutive NF-kappa-B maintains high expression of a characteristic gene network, including CD40, CD86, and a set of antiapoptotic genes in Hodgkin/Reed-Sternberg cells. Blood 97:2798-2807, 2001 19. ten Berge RL, Oudejans JJ, Dukers DF, et al: Percentage of activated cytotoxic T-lymphocytes in anaplastic large cell lymphoma and Hodgkin's disease: An independent biological prognostic marker. Leukemia 15:458-464, 2001[CrossRef][Medline]
20. Oudejans JJ, Jiwa NM, Kummer JA, et al: Activated cytotoxic T cells as prognostic marker in Hodgkin's disease. Blood 89:1376-1382, 1997
21. von Wasielewski R, Seth S, Franklin J, et al: Tissue eosinophilia correlates strongly with poor prognosis in nodular sclerosing Hodgkin's disease, allowing for known prognostic factors. Blood 95:1207-1213, 2000 22. Glavina-Durdov M, Jakic-Razumovic J, Capkun V, et al: Assessment of the prognostic impact of the Epstein-Barr virus-encoded latent membrane protein-1 expression in Hodgkin's disease. Br J Cancer 84:1227-1234, 2001[CrossRef][Medline] 23. Montalban C, Abraira V, Morente M, et al: Epstein-Barr virus-latent membrane protein 1 expression has a favorable influence in the outcome of patients with Hodgkin's disease treated with chemotherapy. Leuk Lymphoma 39:563-572, 2000[Medline]
24. Morente MM, Piris MA, Abraira V, et al: Adverse clinical outcome in Hodgkin's disease is associated with loss of retinoblastoma protein expression, high Ki67 proliferation index, and absence of Epstein-Barr virus-latent membrane protein 1 expression. Blood 90:2429-2436, 1997
25. Kallioniemi OP, Wagner U, Kononen J, et al: Tissue microarray technology for high-throughput molecular profiling of cancer. Hum Mol Genet 10:657-662, 2001 26. Tzankov A, Zimpfer A, Lugli A, et al: High-throughput tissue microarray analysis of G1-cyclin alterations in classical Hodgkin's lymphoma indicates overexpression of cyclin E1. J Pathol 199:201-207, 2003[CrossRef][Medline] 27. Hedvat CV, Hegde A, Chaganti RS, et al: Application of tissue microarray technology to the study of non-Hodgkin's and Hodgkin's lymphoma. Hum Pathol 33:968-974, 2002[CrossRef][Medline]
28. Wang TH, Donaldson YK, Brettle RP, et al: Identification of shared populations of human immunodeficiency virus type 1 infecting microglia and tissue macrophages outside the central nervous system. J Virol 75:11686-11699, 2001
29. Ferme C, Mounier N, Divine M, et al: Intensive salvage therapy with high-dose chemotherapy for patients with advanced Hodgkin's disease in relapse or failure after initial chemotherapy: Results of the Groupe d'Etudes des Lymphomes de l'Adulte H89 Trial. J Clin Oncol 20:467-475, 2002
30. Achten R, Verhoef G, Vanuytsel L, et al: T-cell/histiocyte-rich large B-cell lymphoma: A distinct clinicopathologic entity. J Clin Oncol 20:1269-1277, 2002 31. Longo DL, Duffey PL, Young RC, et al: Conventional-dose salvage combination chemotherapy in patients relapsing with Hodgkin's disease after combination chemotherapy: The low probability for cure. J Clin Oncol 10:210-218, 1992[Abstract] 32. Carde P, Koscielny S, Franklin J, et al: Early response to chemotherapy: A surrogate for final outcome of Hodgkin's disease patients that should influence initial treatment length and intensity? Ann Oncol 13:86-91, 2002[Abstract]
33. Bonfante V, Santoro A, Viviani S, et al: Outcome of patients with Hodgkin's disease failing after primary MOPP- ABVD. J Clin Oncol 15:528-534, 1997
34. DeVita VT Jr, Hubbard SM: Hodgkin's disease. N Engl J Med 328:560-565, 1993 35. Devilard E, Bertucci F, Trempat P, et al: Gene expression profiling defines molecular subtypes of classical Hodgkin's disease. Oncogene 21:3095-3102, 2002[CrossRef][Medline] 36. Hosmer DW, Lemeshow S: Applied Logistic Regression. New York, NY, John Wiley & Sons, 1989 37. Belsley D: Conditioning Diagnostics: Colinearity and Weak Data in Regression. New York, NY, John Wiley & Sons, 1991 38. Bautista D, Arana E, Marti-Bonmati L, et al: Validation of logistic regression models in small samples: Application to calvarial lesions diagnosis. J Clin Epidemiol 52:237-241, 1999[CrossRef][Medline]
39. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143:29-36, 1982 40. Marubini E, Valsecchi MG: Analysing Survival Data From Clinical Trials and Observational Studies. Chichester, United Kingdom, John Wiley & Sons, 1995 41. Canellos GP: Treatment of relapsed Hodgkin's disease: Strategies and prognostic factors. Ann Oncol 9:S91-S96, 1998 42. Josting A, Wolf J, Diehl V: Hodgkin disease: Prognostic factors and treatment strategies. Curr Opin Oncol 12:403-411, 2000[CrossRef][Medline]
43. Axdorph U, Sjoberg J, Grimfors G, et al: Biological markers may add to prediction of outcome achieved by the international prognostic score in Hodgkin's disease. Ann Oncol 11:1405-1411, 2000
44. Josting A, Franklin J, May M, et al: New prognostic score based on treatment outcome of patients with relapsed Hodgkin's lymphoma registered in the database of the German Hodgkin's lymphoma study group. J Clin Oncol 20:221-230, 2002 45. Chronowski GM, Wilder RB, Tucker SL, et al: An elevated serum beta-2-microglobulin level is an adverse prognostic factor for overall survival in patients with early-stage Hodgkin disease. Cancer 95:2534-2538, 2002[CrossRef][Medline]
46. Rassidakis GZ, Medeiros LJ, McDonnell TJ, et al: BAX expression in Hodgkin and Reed-Sternberg cells of Hodgkin's disease: Correlation with clinical outcome. Clin Cancer Res 8:488-493, 2002
47. Rassidakis GZ, Medeiros LJ, Vassilakopoulos TP, et al: BCL-2 expression in Hodgkin and Reed-Sternberg cells of classical Hodgkin disease predicts a poorer prognosis in patients treated with ABVD or equivalent regimens. Blood 100:3935-3941, 2002 48. Benharroch D, Levy A, Prinsloo I, et al: Apoptotic index as a prognostic factor in Hodgkin's disease. Leuk Lymphoma 33:351-359, 1999[Medline] 49. Smolewski P, Niewiadomska H, Los E, et al: Spontaneous apoptosis of Reed-Sternberg and Hodgkin cells; clinical and pathological implications in patients with Hodgkin's disease. Int J Oncol 17:603-609, 2000[Medline] 50. Stokke T, Galteland E, Holte H, et al: Oncogenic aberrations in the p53 pathway are associated with a high S phase fraction and poor patient survival in B-cell non-Hodgkin's lymphoma. Int J Cancer 89:313-324, 2000[CrossRef][Medline]
51. Sanchez-Beato M, Sanchez-Aguilera A, Piris MA: Cell cycle deregulation in B-cell lymphomas. Blood 101:1220-1235, 2003
52. Montesinos-Rongen M, Roers A, Kuppers R, et al: Mutation of the p53 gene is not a typical feature of Hodgkin and Reed-Sternberg cells in Hodgkin's disease. Blood 94:1755-1760, 1999 53. Elenitoba-Johnson KS, Medeiros LJ, Khorsand J, et al: P53 expression in Reed-Sternberg cells does not correlate with gene mutations in Hodgkin's disease. Am J Clin Pathol 106:728-738, 1996[Medline]
54. Garcia JF, Villuendas R, Sanchez-Beato M, et al: Nucleolar p14(ARF) overexpression in Reed-Sternberg cells in Hodgkin's lymphoma: Absence of p14(ARF)/Hdm2 complexes is associated with expression of alternatively spliced Hdm2 transcripts. Am J Pathol 160:569-578, 2002 55. Smolewski P, Niewiadomska H, Krykowski E, et al: Expression of p21 and MDM-2 proteins on tumor cells in responding and non-responding patients with Hodgkin's disease. Neoplasma 46:212-218, 1999[Medline]
56. Dukers DF, Meijer CJ, ten Berge RL, et al: High numbers of active caspase 3-positive Reed-Sternberg cells in pretreatment biopsy specimens of patients with Hodgkin disease predict favorable clinical outcome. Blood 100:36-42, 2002 Submitted June 24, 2003; accepted February 9, 2004. This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
-B), Bax, and STATs; and reveal multiple alterations in different pathways and checkpoints, including G1/S and G2/M transition, and apoptosis control. 
