Leukemia (2010) 24, & 2010 Macmillan Publishers Limited All rights reserved /10.

Transcription

1 ORIGINAL ARTICLE (2010) 24, & 2010 Macmillan Publishers Limited All rights reserved /10 (R-CHOP) is a risk factor for predicting relapse in patients with diffuse large B-cell lymphoma LF Porrata, K Rsitow, DJ Inwards, SM Ansell, IN Micallef, PB Johnston, TM Habermann, TE Witzig, JP Colgan, GS Nowakowski, CA Thompson and SN Markovic Division of Hematology, Department of Medicine, Mayo Clinic College of Medicine, Rochester, MN, USA A specific predictor during routine follow-up to ascertain risk for relapse after standard chemotherapy in non-hodgkin s lymphoma (NHL) has not been identified. Thus, we studied absolute lymphocyte count (ALC) as a marker of poststandard chemotherapy (rituximab, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP)) NHL relapse in patients with diffuse large B-cell lymphoma (DLBCL). ALC was obtained at the time of confirmed relapse and at last follow-up. From 2000 until 2006, 149 consecutive DLBCL patients, originally diagnosed, treated with R-CHOP and followed up at Mayo Clinic, Rochester, were included in this study. Patients at last followup without relapse (N ¼ 112) had a higher ALC compared with those with relapsed lymphoma ((N ¼ 37) median ALC 10 9 /l of 1.43 (range: ) versus 0.67 (range: ), Po0.0001, respectively). ALC at the time of confirmed relapse was a strong predictor for relapse with an area under the curve ¼ 0.91 (Po0.0001). An ALC o /l at the time of confirmed relapse had a positive predictive value of 72% and a positive likelihood ratio of 7.4 to predict relapse after R-CHOP in DLBCL. Patients with an ALCX /l (N ¼ 103) had a cumulative incidence of relapse of 6 versus 79% with an ALC o /l (N ¼ 46) (Po0.0001). This study suggests that lymphopenia measured by ALC can be used as a marker to assess risk of DLBCL relapse during routine follow-up after standard chemotherapy. (2010) 24, ; doi: /leu ; published online 20 May 2010 Keywords: absolute lymphocyte count; diffuse large B-cell lymphoma; relapse Introduction The current recommendation for long-term follow-up in patients with non-hodgkin s lymphoma (NHL), who are not participating in clinical trials, by the National Comprehensive Cancer Network is a follow-up every 3 6 months or as clinically indicated. 1 The European Society of Medical Oncology s recommendations for follow-up in NHL include follow-up every 3 months for 2 years, every 6 months for 3 years and then once a year. 2 The European Society of Medical Oncology 2 also recommended blood counts and lactate dehydrogenase (LDH) at 3, 6, 12 and 24 months, then only as needed for evaluation of suspicious symptoms or clinical finding in those patients suitable for further therapy. Regarding imaging studies, the European Society of Medical Oncology 2 recommended minimal adequate radiological examinations at 6, 12 and 24 months after the end of treatment by computed tomography (CT) Correspondence: Dr LF Porrata, Division of Hematology, Department of Medicine, Mayo Clinic College of Medicine, 200 First St SW, Rochester, MN, 55905, USA. porrata.luis@mayo.edu Received 9 March 2010; accepted 30 March 2010; published online 20 May 2010 scans when indicated by site of disease. There is no evidence to show that scheduled surveillance CT scans or positron emission tomography improves clinical outcomes and the patient and physician can identify relapse more than 80% of the time without the need for imaging studies. 3,4 Therefore, the utilization of radiological examinations has been left to the discretion of the physician to use as needed to evaluate suspicious symptoms or clinical findings that suggest relapse. 3,4 However, complete blood cell count and serum chemistry, including LDH, have been recommended for each follow-up visit. 3,4 Risk factors used to assess clinical outcomes in NHL patients treated with standard therapy are identified before treatment implementation, such as the international prognostic index (IPI) 5 or gene expression profiling. 6,7 Even though these risk factors are useful to guide physicians in the identification of patients who would benefit from standard therapy, a limitation of these risk factors is that they are tested at one point in time. An inexpensive factor that could be monitored during follow-up after standard therapy and predicts relapse would be a powerful tool to help clinicians identify patients who might require further evaluation for relapse. In NHL, LDH has been reported to have a sensitivity of 42% and a specificity of 85% to predict relapse during follow-up in diffuse large B-cell lymphoma (DLBCL). 8 Recently, absolute lymphocyte count (ALC) at diagnosis, as a surrogate marker of host immunity, has been reported to be a prognostic factor for clinical outcomes in DLBCL Thus, we set out to investigate whether ALC at the time of confirmed relapse or at last follow-up, obtained from a complete blood cell count as recommended in follow-up for DLBCL, is a marker for relapse after standard therapy in DLBCL. Materials and methods Patient population To participate in this study, patients were required to have a diagnosis of de novo DLBCL; be treated with rituximab, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) with or without consolidation radiation therapy at the discretion of the attending physician; and be followed up at Mayo Clinic, Rochester. Patients with primary DLBCL central nervous system lymphoma, transformed NHL, postsolid organ transplant lymphoproliferative disorder, positive human immunodeficiency virus or progressed during R-CHOP were excluded from the study. Patients who underwent up-front autologous stem cell transplantation after R-CHOP were also excluded. From 20 December 2000 until 28 July 2006, 149 DLBCL patients qualified for the study. Data from DLBCL patients were collected prospectively and entered into a computerized database. No patients were lost to follow-up. All patients gave written, informed consent allowing the use of their medical

2 1344 records for medical research. Approval for the retrospective review of these records was obtained from the Mayo Clinic Institutional Review Board and was in accordance with US federal regulations and the Declaration of Helsinki. variables. The Mahalanobis distance was used as an independent approach to assess the robustness of the Pearson correlation. All P values represented were two sided and statistical significance was declared at Po0.05. End point The primary end point of the study was to assess whether ALC at the time of confirmed relapse or at last follow-up is a reliable marker to predict relapse after R-CHOP in DLBCL. The ALC was obtained from the complete blood cell count. 12 Risk factors for relapse Risk factors tested in the study included ALC at the time of confirmed relapse or at last follow-up, LDH at the time of confirmed relapse or at last follow-up, IPI 3 (agex60, extranodal sites X2, LDH (abnormal versus normal levels), performance status X2 and stage (I/II versus III/IV)), postchemotherapy consolidation radiation and ALC at diagnosis (ALC at diagnosis X800 cells/ml). This cutoff value for the ALC at diagnosis was based on data from reported studies. 9,10 Chemotherapy All patients received rituximab 375 mg/m 2, cyclophosphamide 750 mg/m 2, doxorubicin 50 mg/m 2, vincristine 1.4 mg/m 2 and prednisone 100 mg/m 2 5 days. Response Response criteria were based on criteria from the Lymphoma International Workshop. 1 Relapse was defined as any new lesions or increase by X50% of previously involved sites from nadir. 7 The time to relapse was measured from the date of completed treatment to relapse. Last follow-up was measured from the date of completed treatment to the day of last follow-up or death in patients without any evidence of relapse. Statistical analysis To assess the effect of ALC at the time of confirmed relapse or at last follow-up on relapse after R-CHOP, the relapse end point was examined both as cumulative incidence and cumulative hazard function plot. The cumulative incidence explicitly accounts for death from other causes besides lymphoma as competing risks and was estimated using Gooley s method. 13 The hazard function is the principal estimable quantity in competing risks, which can be viewed as a probability of failure specifically resulting from a cause in a small interval of time, given that no failure of any kind has occurred thus far. The cumulative hazard function L (t) equals the value of its corresponding hazard function summed up to time t. 14 The cumulative hazard function was calculated using the Nelson Aalen estimator. 15 The association between ALC at the time of confirmed relapse or at last follow-up and risk factors with the incidence of relapse was also explored using logistic regression models. The choice of optimal cutoff of ALC at the time of confirmed relapse or at last follow-up was based on its utility as a marker for relapse using box plot, receiver operating characteristics curves and area under the curve. w 2 -tests were used to determine relationships between categorical variables. The Wilcoxon-rank test was used to determine associations between continuous variables and categories, and Pearson s correlation coefficients were used to evaluate associations for continuous Results Patient characteristics The median age at the time of diagnosis for this cohort of 149 DLBCL patients was 65 years (range: years). The distribution of additional baseline characteristics for these patients is presented in Table 1 and is summarized on the basis of whether patients had an ALC o /l versus ALC X /l at last follow-up or at the time of confirmed relapse. The only differences between the groups were ALC at diagnosis, bone marrow involvement at diagnosis, IPI, LDH at diagnosis and stage at diagnosis. Despite these differences, we identified no association between ALC at the time of confirmed relapse or at last follow-up and ALC at diagnosis (P ¼ 0.1), bone marrow involvement at diagnosis (P ¼ 0.1), IPI (P ¼ 0.1), LDH at diagnosis (P ¼ 0.7) and stage at diagnosis (P ¼ 0.1). The median follow-up on living patients (N ¼ 107) in this cohort was 41 months (range: ). In all, 37 (25%) patients had confirmed relapse after R-CHOP with a median time to relapse of 10.6 months (range: months). Of the 42 (28%) patient deaths, 26 (62%) were due to relapse of lymphoma and 16 (38%) due to nonlymphoma causes, including 8 patients with heart attack, 2 patients with congestive heart failure, 2 patients with pneumonia, 2 patients with melanoma, 1 patient with brain hemorrhage and 1 patient with acute myelogenous leukemia. ALC at the time of confirmed relapse or at last follow-up after R-CHOP In an attempt to identify risk factors during follow-up that predict relapse after R-CHOP, we assessed the utility of ALC at the time of confirmed relapse or at last follow-up after R-CHOP as a marker for this relevant clinical outcome. Despite some overlap, the box plot showed that patients without evidence of relapse (N ¼ 112) at last follow-up had a higher ALC compared with those with relapse (N ¼ 37) after R-CHOP (median ALC 10 9 /l of 1.43 (range: ) versus 0.67 (range: ), Po0.0001, respectively) (Figure 1). Receiver operating characteristics and area under the curve analysis showed that ALC at follow-up was a significant marker for relapse after R-CHOP (area under the curve ¼ 0.91, Po0.0001; Figures 2). On the basis of these results, an ALC cutoff at follow-up of /l was considered optimal. Therefore, this cutoff is evaluated for ALC at follow-up in all subsequent analyses in this study. To assess the predictive value of ALC for relapse at the time of follow-up after R-CHOP, a contingency table was created between ALC X versus o /l at last follow-up and relapse status after R-CHOP (Table 2). Of the 46 patients with an ALC o /l, 71% experienced relapse compared with 3.8% with an ALC X /l (Po0.0001). The sensitivity and specificity for ALC at the time of confirmed relapse or at last follow-up was 89 and 88%, respectively. The relative risk of relapse with an ALC o /l at the time of confirmed relapse or at last follow-up was 18.7 (95% cumulative incidence : ) with an odds ratio of 63.5 (95% cumulative incidence : ). The positive predictive value with an ALC o /l at the time of confirmed relapse or at last follow-up was 72% and the negative predictive value with an

3 Table 1 Patient s baseline characteristics based on ALC o0.96 versus ALC X /l at relapse or at last follow-up 1345 Characteristics ALC o /l (N ¼ 46) ALC X /l (N ¼ 103) P-value Age at diagnosis, years; median (range) 64 (28 86) 56 (19 87) 0.3 Gender 0.8 Female Male Prognostic factors at diagnosis Age, years 0.4 X o Extra-nodal sites p LDH (U/l) o0.01 Normal Abnormal Performance status 0.6 X o Stage o I/II III/IV IPI score o ALC at diagnosis o cells/ml X o Hemoglobin at the time of confirmed relapse or at last follow-up: median (range), g/dl 13.2 ( ) 13.8 ( ) 0.1 WBC at the time of 5.9 ( ) 6.3 ( ) 0.3 confirmed relapse or last follow-up: median (range) 10 9 /l ANC at the time of 4.2 ( ) 3.8 ( ) 0.5 confirmed relapse or last follow-up: median (range) 10 9 /l Plts at the time of confirmed relapse or last follow-up: median 289 (8 688) 270 (38 494) 0.4 (range) 10 9 /l BM at diagnosis o0.003 Positive Negative Post-R-CHOP consolidation xrt Yes 7 29 No Abbreviations: ALC, absolute lymphocyte count; ANC, absolute neutrophil count; BM, bone marrow; IPI, international prognostic index; LDH, lactate dehydrogenase; Plts, platelets; PR, partial remission; R-CHOP, rituximab, cyclophosphamide, adriamycin, vincristine and prednisone; WBC, white blood cell count; xrt, radiation therapy. ALC X /l at the time of confirmed relapse or at last follow-up was 96%. The positive likelihood ratio for relapse with an ALC o /l was 7.4. The negative likelihood 0.1 Figure 1 Box plot of absolute lymphocyte count (ALC) at follow-up post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) in patients with evidence of relapse and patients with no relapse. The horizontal line within each box represents the median and the lower and upper borders of each box represent the 25th and the 75th percentiles, respectively. Outliers (values that exceed these boundaries) are depicted as single points. The median ALC at follow-up post-r-chop was significantly higher in those patients with no relapse versus patients with relapse (1.43 versus /l, respectively, Po ). Figure 2 Receiver operating characteristic (ROC) curve to evaluate absolute lymphocyte count (ALC) at follow-up post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) as a marker for relapse. Specifically, this curve plots the sensitivity versus one minus the specificity of this marker for relapse. The corresponding area under the curve (AUC) analysis indicated that ALC at follow-up after R-CHOP was indeed, a significant marker for relapse (AUC ¼ 0.91, Po0.0001). ratio for relapse with an ALC X /l was 0.1. The crude incidence rate of relapse was 25% (37 of 150). The 5-year cumulative incidence rate for an ALC o /l at the time of confirmed relapse or at last follow-up was 79% compared with 6% for an ALC X /l at the time of confirmed relapse or at last follow-up (Po0.0001) (Figure 3). The 5-year cumulative hazard rate for an ALC o /l at follow-up was 2.1 compared with 0.05 for an ALC X /l at the time of confirmed relapse or at last follow-up (Po0.0001) (Figure 4).

4 1346 Table 2 Count Contingency table ALC o /l at the time of confirmed relapse or last follow-up post-r-chop ALC X /l at the time of confirmed relapse or last follow-up post-r-chop Relapse at follow-up after R-CHOP Yes No 33 (a) 13 (b) 46 4 (c) 99 (d) Abbreviations: ALC, absolute lymphocyte count; R-CHOP, rituximab, cyclophosphamide, adriamycin, vincristine and prednisone. Formulas: sensitivity ¼ a/(a+c); specificity ¼ d/(b+d); positive predictive value ¼ a/(a+b); negative predictive value ¼ d/(c+d); positive likelihood ratio ¼ sensitivity/(1 specificity); negative likelihood ratio ¼ (1 sensitivity)/ specificity; odd ratio ¼ ad/bc; relative risk ¼ (a/(a+b))/(c/(c+d)). Univariate and multivariate logistic regression Logistic regression models for predicting relapse after R-CHOP further indicate that ALC at the time of confirmed relapse or at last follow-up is significantly correlated with this clinical outcome (Po0.0001), with an odds ratio of 63.4 (Table 3). Other significant factors for relapse after R-CHOP in the univariate setting included ALC at diagnosis, elevated LDH at diagnosis and at the time of confirmed relapse or at last follow-up, IPI, post-r-chop consolidation radiation therapy and stage. When these factors were accounted for in addition to ALC at the time of confirmed relapse or at last follow-up in a multivariate logistic regression model (Table 4), ALC at the time of confirmed relapse or at last follow-up remained a significant correlate for relapse after R-CHOP (Po0.0001). An ALC o /l at the time of confirmed relapse or at last follow-up was associated with an adjusted odds ratio for relapse after R-CHOP of 42.3 (95% cumulative incidence : ). The LDH at the time of confirmed relapse or at last follow-up was nearly statistically significant (P ¼ 0.07). Figure 3 Cumulative incidence (CI) for relapse based on absolute lymphocyte count (ALC) at follow-up post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) Patients with an ALC o /l at follow-up experienced a higher CI of 79% compared with a CI of 6% for patients with an ALC X /l at follow-up after R-CHOP. Figure 4 Cumulative hazard rate for relapse based on absolute lymphocyte count (ALC) at follow-up post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP) Patients with an ALC o /l at follow-up had a cumulative hazard rate of 2.1 compared with only 0.05 for patients with an ALC X /l at follow-up APHSCT. Table 3 Logistic regression univariate analysis for relapse after R-CHOP Variable Estimate s.e. w 2 OR P-value ALC o at diagnosis o ALC o at relapse or at follow-up after R-CHOP o Age 460 years Gender Extra nodal sites LDH at diagnosis o0.001 LDH at relapse or at follow-up after R-CHOP o PS Stage III/IV o IPI X o Post R-CHOP consolidation radiation therapy o0.04 Abbreviations: ALC, absolute lymphocyte count; IPI, international prognostic index; LDH, lactate dehydrogenase; OR, odds ratio; PS, performance status; R-CHOP, rituximab, cyclophosphamide, adriamycin, vincristine and prednisone; w 2, chi square. The degree of freedom for the statistical analysis was 1.

5 Table 4 Logistic regression multivariate analysis for relapse after R-CHOP 1347 Variable Estimate s.e. w 2 OR P-value ALC o at diagnosis ALC o at relapse or at follow-up after R-CHOP o LDH at diagnosis LDH at relapse or follow-up after R-CHOP Stage III/IV IPI X Post R-CHOP consolidation radiation therapy Abbreviations: ALC, absolute lymphocyte count; IPI, international prognostic index; LDH, lactate dehydrogenase; OR, odds ratio; PS, performance status; R-CHOP, rituximab, cyclophosphamide, adriamycin, vincristine, and prednisone; w 2, chi square. The degree of freedom for the statistical analysis was 1. Figure 5 Scatter plot comparing absolute lymphocyte count (ALC) at follow-up and lactate dehydrogenase (LDH) at follow-up post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP). A negative correlation was identified between ALC and LDH at follow-up after APHSCT. Arrows indicate those points that according to the Mahalanobis distances are outliers. R1 and R2 correspond to the Pearson s r values before and after eliminating possible outliers. The regression line was estimated after the elimination of outliers. We identified no association between post-r-chop consolidation radiation therapy and ALC at the time of confirmed relapse or at last follow-up (P ¼ 0.9). In addition, there was no difference between the groups (ALC X0.96 versus ALC o0.96) regarding hemoglobin, white blood cell count, absolute neutrophil count or platelet count obtained at last follow-up or at the time of confirmed relapse (Table 1). Of the 37 patients with relapse, 3 (8%) had positive bone marrow involvement, all of whom had an ALC o /l at the time of confirmed relapse. Detection of relapse Of the 37 cases with proven relapsed disease, further evaluation for relapse was carried out in 21 (57%) cases because of patient s symptoms, 4 (11%) cases because of physical findings during routine physical examination and 12 (32%) cases because of CT or positron emission tomography scan findings ordered by the attending physician. Thus, in 68% of cases, relapse was suggested by the combination of patient s symptoms and physical findings on routine physical examination. In the 12 cases in which patients presented without any symptoms or physical findings and relapse was identified by CT or positron emission tomography scans, 10 of 12 (87%) cases had an ALC o /l and 4 of 12 (33%) cases had an abnormal LDH. Figure 6 Cumulative incidence (CI) for relapse based on the absolute lymphocyte count (ALC) and lactate dehydrogenase (LDH) at followup post-rituxan, cyclophosphamide, adriamycin, vincristine and prednisone (R-CHOP). Patients with an ALC o /l experienced a higher CI of relapse regardless whether the LDH was normal or abnormal at follow-up compared with a lower CI of relapse with an ALC X /l regardless whether the LDH was normal or abnormal at follow-up post-r-chop. P-values for high ALC/normal LDH versus high ALC/abnormal LDH was 0.1, high ALC/normal LDH versus low ALC/normal LDH was o0.0001, high ALC/normal LDH versus low ALC/abnormal LDH was o0.0001, high ALC/abnormal LDH versus low ALC/normal LDH was o0.001, high ALC/abnormal LDH versus low ALC/abnormal LDH was o and low ALC/normal LDH versus low ALC/abnormal LDH was ALC and LDH at the time of confirmed relapse or at last follow-up after R-CHOP LDH at the time of confirmed relapse or at last follow-up after R-CHOP in this study showed a statistical trend as a risk factor for relapse after R-CHOP. LDH is considered as a surrogate marker for tumor burden in NHL. Similarly, ALC can be considered as a surrogate marker of host immunity. 11 Thus, using these two standardized, low-cost biomarkers, we assessed whether there was any correlation between tumor burden and host immunity. We identified a negative correlation between ALC and LDH before (r ¼ 0.36, Po0.0001) and after (r ¼ 0.32, Po0.0001) outliers identified by the Mahalanobis distance were eliminated (Figure 5). As a result of this negative correlation and the fact that both markers are associated with relapse when checked during follow-up after R-CHOP, we assessed the cumulative incidence of relapse combining both markers. We found a lower cumulative incidence of relapse in patients with a higher ALC

6 1348 compared with patients with a lower ALC, regardless of whether LDH was normal or abnormal at follow-up after R-CHOP (Figure 6). Discussion The current risk factors used to assess prognosis in NHL patients treated with standard therapy are identified before treatment However, a limitation of these risk factors is that they are tested at one point in time and they are assumed to retain their predictive ability to determine clinical outcome at any time after standard therapy. An ideal risk factor is a risk factor that has the ability to predict not only future clinical outcomes but also clinical outcome at any given time point after therapy has been implemented. Therefore, we set out to investigate whether ALC checked at any given time during follow-up is a marker of relapse in DLBCL patients treated with R-CHOP. This study shows that DLBCL patients with confirmed relapse at follow-up had a lower ALC compared with those without evidence of relapse at last follow-up after R-CHOP. A low ALC at the time of confirmed relapse or at last follow-up was associated with a high odds ratio, high relative risk, high sensitivity, high positive predictive value, high cumulative hazard rate and high cumulative incidence for relapse compared with a high ALC at the time of confirmed relapse or at last follow-up, which was associated with a high specificity, high negative predictive value and low cumulative hazard rate and cumulative incidence for relapse after R-CHOP. In accordance with the literature, in 68% of the relapsed cases, 3 the detection of relapse was triggered by patients symptoms or physical examination findings. Of the 12 cases in which unsuspected relapse was detected by CT or positron emission tomography scan, 87% of the cases had a low ALC and only 33% had an abnormal LDH. These results suggest that DLBCL patients with no symptoms or physical findings and a low ALC at follow-up may require imaging to detect relapses. The association of decreased ALC and increased risk of NHL relapse could be the result of either (a) primary failure of immune surveillance yielding a permissive systemic immunological environment allowing for clinical NHL relapse or (b) primary NHL relapse driven by tumor-associated events that in turn produce mediators of immune suppression manifesting as a decrease in ALC. The data presented in this study are unable to distinguish between these two possibilities. However, to the extent that ALC can be viewed as a surrogate marker of immune competence, there are numerous examples in which the presence of competent systemic immunity directly affects NHL biology. Possibly the best clinical example in support of this notion is the natural history of posttransplant lymphoproliferative disorders. 16 In this context, a frequently effective therapeutic intervention against NHL is a simple reduction of immunosuppression, currently used as standard of care. An associated, although less-profound clinical observation, is the association of ALC and efficacy of rituximab therapy in patients with low-grade NHL. 17 Patients with normal ALC have a significantly higher likelihood of good clinical outcomes following rituximab therapy relative to those with low ALC. Assuming that one aspect of effective rituximab therapy is its ability to mediate antibody-dependent cellular cytotoxicity, presence of competent immunity (reflected by normal ALC) seems to be critical to clinical outcomes. However, despite these associations, a mechanistic explanation for the increased risk for NHL in patients who show low ALC can only be addressed in an appropriately designed prospective clinical trial in which relevant analyses of both systemic immunity and tumor phenotype can be studied. Such an endeavor is currently under way. Nevertheless, on the basis of presented data, the association between ALC and NHL relapse seems to be clinically useful in judging risk for NHL relapse in patients in clinical follow-up after R-CHOP. To minimize the inherent biases of a retrospective study, the following steps were taken. First, we selected only patients with de novo DLBCL and excluded other presentations of DLBCL such as primary central nervous system lymphoma, posttransplant lymphoproliferative disorder or transformed NHL, because these presentations of DLBCL are not treated uniformly. Second, we selected patients with de novo DLBCL, because de novo DLBCL is the most common lymphoma type for which R-CHOP is indicated as the standard of care, regardless of prognostic factors such as IPI or gene expression profiling. 18,19 Third, we could not identify any medications, infections or post-r-chop consolidation radiation therapy that could have affected the ALC at follow-up. The strengths of this study included the analysis of a large cohort of uniformly treated sequential patients with prolonged follow-up. This study identifies a worldwide, standardized and low-cost risk factor to assess risk of relapse after R-CHOP. To our knowledge, this study is the first to identify ALC at follow-up as a marker to assess relapse at any given point in time during follow-up after R-CHOP in DLBCL. Thus, our study suggests that the ALC at follow-up can be used as a simple, inexpensive tool to alert clinicians of relapse during follow-up after R-CHOP in DLBCL. Conflict of interest The authors declare no conflict of interest. References 1 National Comprehensive Cancer Network (NCCN). Diffuse large B-cell lymphoma. Practice Guidelines in Oncology-v Tilly H, Dreying M. Diffuse large B-cell non-hodgkin s lymphoma: ESMO clinical recommendations for diagnosis, treatment and follow-up. Ann Oncol 2009; 20 (suppl 4): iv110 iv112: doi: /annonc/mdp Cheson BD, Pfistner B, Juweid ME, Gascoyne RD, Specht L, Horning SJ et al. Revised response criteria for malignant lymphoma. J Clin Oncol 2007; 25: Schoulten HC. Response evaluation and long term follow-up. Ann Oncol 2007; 18 (Suppl 1): i33 i36. 5 Shipp MA, Harrington DP, Anderson JR, Armitage JO, Bonadonna G, Brittinger G et al. A predictive model for aggressive NHL: The International non-hodgkin s Lymphoma Prognostic Factors Project. N Engl J Med 1993; 329: Rosenwald A, Wright G, Chan WC, Connors JM, Campo E, Fisher RJ et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large B-cell lymphoma. N Engl J Med 2002; 346: Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000; 403: Weeks JC, Yeap BY, Canellos GP, Shipp MA. Value of follow-up procedures in patients with large-cell lymphoma who achieves a complete remission. J Clin Oncol 1991; 9: Cox MC, Nofroni I, LaVerde G, Ferrari A, Amodeo R, Tatarelli C et al. Absolute lymphocyte count is a prognostic factor for diffuse large B-cell lymphoma. Br J Haematol 2008; 141: Cox MC, Nofroni I, Ruco L, Amodeo R, Ferrari A, La Verde G et al. Low absolute lymphocyte count is a poor prognostic factor in diffuse large B-cell lymphoma. Leuk Lymphoma 2008; 49: Oki Y, Yamamoto K, Kato H, Kuwatsula Y, Taji H, Kagami Y et al. Low absolute lymphocyte count is a poor prognostic marker

7 in patients with diffuse large B-cell lymphoma and suggests patients survival benefit from rituximab. Eur J Haematol 2008; 81: Cox CJ, Habermann TM, Payne BA, Klee GG, Pierre RV. Evaluation of the Coulter counter model S-Plus IV. Am J Clin Pathol 1985; 84: Gooley TA, Leisenring W, Crowley J, Storer BE. Estimation of failure probabilities in the presence of competing risks: new representation of old estimators. Stat Med 1999; 18: Dignam JJ, Kocherginsky MN. Choice and interpretation of statistical tests used when competing risks are present. J Clin Oncol 2008; 26: Ghosh D. Proportional hazards regression for cancer studies. Biometrics 2008; 64: Habermann TM. Posttransplant lymphoproliferative disorders. Cancer Treat Res 2008; 142: Behl D, Ristow K, Markovic SN, Witzig TE, Habermann TM, Colgan JP et al. Absolute lymphocyte count predicts therapeutic efficacy of rituximab therapy in follicular lymphomas. Br J Haematol 2007; 137: Coiffier B, Lepage E, Briere J, Herbrecht R, Tilly H, Bouabdallah R et al. CHOP chemotherapy plus rituximab compared with CHOP alone in elderly patients with diffuse large B-cell lymphoma. N Engl J Med 2002; 346: Habermann TM, Weler EA, Morrison VA, Gascoyne RD, Cassileth PA, Cohn JB et al. Rituximab-CHOP versus CHOP alone or with maintenance rituximab in older patients with diffuse large B-cell lymphoma. J Clin Oncol 2006; 24: