Article Text
PDF
Abstract
Background Bone marrow involvement confers a poor prognosis in patients with diffuse, large, B-cell lymphoma (DLBCL). However, the prognostic significance of concordant and discordant bone marrow involvement in these cases differs. We analysed this further in patients treated with R-CHOP (rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone) at a single institute.
Design and Methods The cytomorphology of bone marrow involvement was evaluated in 632 patients who were diagnosed with DLBCL in primary tissues and had received R-CHOP therapy. Bone marrow trephine biopsies and clot sections were analysed, along with the immunohistochemical analysis of CD20, CD79a and CD3.
Results Bone marrow involvement was identified in 80 of our DLBCL patient subjects (12.7%). Of these, 32 (40%) showed discordant bone marrow involvement, and 48 (60%) showed concordant involvement. Kaplan–Meier survival analysis showed that progression-free survival and overall survival was poorer in the concordant group (p<0.001). Multivariate analysis, adjusted for the International Prognostic Index score, showed that concordant involvement was an independent predictor of progression-free survival (p<0.001) and overall survival (p=0.011). Discordant involvement was not a negative prognostic factor independent of the International Prognostic Index.
Conclusions Prognostication based on bone marrow involvement cytomorphology is a useful indicator of progression-free survival and overall survival, independent of the International Prognostic Index score, in DLBCL patients. Accurate staging based on morphology should thus be included in bone marrow examinations of such cases.
- Immunohistochemistry
- Lymphoma
- Bone Marrow
Statistics from Altmetric.com
Introduction
Diffuse, large, B-cell lymphoma (DLBCL) is the most common form of non-Hodgkin lymphoma.1 Bone marrow (BM) involvement has been reported in 11–27% of these cases.2 ,3 Although BM involvement may mostly be DLBCL (concordant), discordant morphological variants, which comprise small B-lymphocytes, may also be present. Concordant BM involvement is associated with a poor prognosis in these cases,3 but the impact of discordant involvement remains unclear. In patients with aggressive non-Hodgkin lymphoma, the International Prognostic Index (IPI) is a powerful tool for predicting clinical outcome. The IPI, which is based on age, stage, extranodal site involvement, lactate dehydrogenase levels and performance status,4 must be considered whenever morphological, molecular and immunophenotypic subgroups are evaluated as prognostic factors.
Since the introduction of rituximab as a treatment for DLBCL, both outcomes and previously recognised prognostic factors have changed.5 ,6 In our present study, the prognostic impact of concordant and discordant BM involvement was evaluated in DLBCL patients treated with R-CHOP (rituximab plus cyclophosphamide, doxorubicin, vincristine and prednisone) at a single institute. To determine its independent effect, the impact of the morphological/histological subtype of BM involvement was corrected for the IPI.
Design and methods
Patient samples and methods
We included all patients with electronic medical records at the Asan Medical Center who met the following criteria: (1) a confirmed diagnosis of DLBCL on pathology review; (2) treatment with rituximab immunotherapy and combination therapy (R-CHOP); (3) BM examination and (4) in cases of BM involvement, availability of the BM slides for review.
All patients were staged according to the Ann Arbor system.7 Performance status was assigned according to the Eastern Cooperative Oncology Group (ECOG) scale,8 and the IPI was calculated as described previously.4
BM trephine biopsies and clot sections were reviewed by two haematopathologists, who recorded morphological and histological subtypes as either negative or positive, with either concordant or discordant BM involvement. The first screen for BM involvement involved examination of sections stained with H&E. Suspected involvement was confirmed by immunohistochemical analyses using monoclonal antibodies specific for CD20 (Novocastra, Newcastle, UK), CD3 (DAKO, Denmark), and CD79a (DAKO, Glostrup, Denmark), with routine protocols for automated immunohistochemistry involving the Ventana Benchmark XT instrument (Ventana Medical Systems, Tucson, Arizona, USA). Concordant involvement was defined by the involved area consisting mostly of large cells with prominent nucleoli and cytoplasm (figure 1). Discordant involvement was defined by the involved area consisting mostly of small cells forming lymphoid aggregates (figure 2), and was distinguished from reactive lymphocyte aggregates by definite CD20 positivity without infiltration of CD3-positive lymphocytes.
Images of concordant involvement. (a) H&E stain×400. (b) CD20×400. (c) CD3×400. (d) H&E stain of original tumour×400. This figure is only reproduced in colour in the online version.
Images of discordant involvement. (a) H&E stain×200. (b) CD20×200. (c) CD3×200. (d) H&E stain of original tumour×400. This figure is only reproduced in colour in the online version.
In addition to biopsy trephines, BM particle sections (‘clot sections’) were used to increase the sensitivity with which BM involvement was detected. We routinely collected approximately 0.5–1.0 ml of aspirated BM in 50 ml of 10% neutral formalin fixative and incubated the mixture for 4 h at 56–60°C. The particles were then placed in 70% alcohol to induce cell lysis. BM particles were placed on lens paper, processed through alcohol and xylene, and embedded in paraffin.
Ethical approval for this study was obtained from the institutional review board of Asan Medical Center, which is officially accredited by the Forum for Ethical Review Committees in Asia and the Western Pacific.
Statistical analysis
To compare clinical characteristics between the subject groups, the independent t test was used for continuous variables, and the χ2 test was used for categorical variables. Progression-free survival (PFS) was calculated from the date of first pathological diagnosis to the time of relapse, disease progression or death from any cause. Disease progression was defined by clinical features, and relapse was defined by both clinical and pathological diagnosis. Overall survival (OS) was calculated as the time from the date of first pathological diagnosis until death from any cause. The Kaplan–Meier method was used to assess PFS and OS, and the log-rank test was used for comparisons between groups. Kaplan–Meier analysis was used for PFS and OS according to IPI score (figure 3). The Cox proportional hazard model was used for multivariate analysis to correct for the effect of IPI and assess the independent effects of prognostic variables on outcomes. Data were analysed using SPSS software (SPSS V.18.0 for Windows; SPSS Inc, Chicago, Illinois, USA). BM involvement is included in the IPI score as extranodal involvement. To compare the predictive value of relapse and death, a 2×2 table was drawn for concordant versus non-concordant, and extranodal versus non-extranodal involvement. This was used to determine the sensitivity, specificity, positive predictive value and negative predictive value.
Kaplan–Meier plot of progression-free survival (PFS) and overall survival according to each International Prognostic Index score. (A) PFS (p=0.000). (B) PFS (p=0.000). This figure is only reproduced in colour in the online version.
Results
Patient characteristics
Six hundred and thirty-two patients met the inclusion criteria. Eighty patients (12.7%) presented with BM involvement at the time of diagnosis, whereas, 32 patients (40%) had discordant BM involvement, and 48 patients (60%) had concordant involvement. The baseline clinical characteristics associated with discordant, concordant or negative BM involvement are listed in table 1. The median age was significantly different between the three groups (p <0.001), as were all the IPI factors (p <0.001), except for ECOG performance status. When the Kaplan–Meier analysis was carried out between two IPI scores, results of the log-rank sum test between the IPI scores 0 and 1 were: p=0.143 for PFS and p=0.102 for OS; between IPI scores of 1 and 2, p=0.085 for PFS and p=0.130 for OS; between IPI scores of 2 and 3, p=0.007 for PFS and p=0.001 for OS; between IPI scores of 3 and 4, p=0.047 for PFS and p=0.099 for OS; between IPI scores of 4 and 5, p=0.271 for PFS and p=0.291 for OS. Based on the plot, the IPI score was considered to be low when it was between zero and two, and high when it was between three and five. Owing to these differences, the percentage of patients with a high IPI score in the concordant group (66.7%) was significantly higher (p <0.001) than the percentage of patients in the discordant (40.6%) and negative groups (27.9%).
Patient and clinical characteristics classified according to type of bone marrow involvement
Survival
The median length of follow-up of the cohort at the time of analysis was 24 months (range 1–105 months). Causes of death included pneumonia, brain hemorrhage, septic shock, renal failure, invasive aspergillosis and multiple organ failure. The 3-year PFS and OS rates were significantly different between the groups (PFS: 69.1% vs 57.1% vs 29.6% for the negative, discordant and concordant groups, respectively (p <0.001), and OS: 73.5% vs 64.2% vs 38.5% for the negative, discordant and concordant groups, respectively (p <0.001)) (figure 4). The PFS and OS outcomes between the negative and discordant groups were not significantly different (p=0.30).
Kaplan–Meier survival curves showing progression-free survival (PFS) and overall survival (OS) for the concordant, discordant and negative bone marrow involvement groups. (a) PFS; (b) OS. This figure is only reproduced in colour in the online version.
Multivariate analysis using the Cox proportional hazard model showed that concordant BM involvement had a negative effect, independent of the IPI score, on the PFS (relative risk (RR), 1.98; 95% CI 1.3 to 3.0; p <0.001) and also on OS (RR, 1.76; 95% CI 1.1 to 2.7; p=0.011). Discordant BM involvement had no significant effect either on the PFS (RR, 1.21; 95% CI 0.7 to 2.4; p=0.471) or OS (RR, 1.35; 95% CI 0.7 to 2.8; p=0.415). The effect of the IPI on PFS and OS was higher than that of concordant BM involvement according to multivariate analysis (table 2).
Cox regression analysis of bone marrow involvement, IPI score and individual IPI factors for PFS and OS
Leukaemic manifestation of lymphoma, defined as more than 25% of neoplastic lymphoid cells in the aspirate, was observed in 12 of the patients with BM involvement (12/80, 15%). Ten of these patients were in the concordant group (10/48, 20.8%), and two were in the discordant group (2/32, 6.3%). Leukaemic manifestation was more frequent in the concordant group than in the discordant group (p=0.017). Of the 80 cases showing BM involvement, infiltration of both the trephine and clot sections was more frequently observed in the concordant group (p <0.001). By comparison, the discordant group was more frequently observed in clot sections only (p=0.001).
The specificity and positive predictive value of concordant involvement in predicting death and relapse was higher than extranodal involvement, whereas, sensitivity and negative predictive value were lower for concordant involvement than for extranodal involvement (table 3).
Sensitivity, specificity, positive and negative predictive values of extranodal involvement and concordant bone marrow involvement in predicting death and relapse
Discussion
BM examination for lymphoma is routinely performed to evaluate the stage and extent of the disease at diagnosis. BM is included in the IPI as an extranodal involvement site.4 However, BM involvement can be further classified according to the cytomorphology of the lymphoma involved. In the present study, Kaplan–Meier survival analysis revealed that the concordant group had the worst survival outcomes, whereas, there was no difference between the negative and discordant groups in this regard. Moreover, multivariate analysis via the Cox model revealed that concordant BM involvement is a poor risk factor, independent of the IPI score.
When extranodal involvement and concordant BM involvement were each evaluated for survival (OS) and relapse (PFS) in diagnostic tests (2×2 table), the latter showed higher specificity and positive predictive value and lower sensitivity in both survival and relapse. Given that the patients are already diagnosed with DLBCL on the basis of tissue pathology, the positive predictive value for relapse and death is the appropriate diagnostic characteristic of concordant BM involvement when assessing prognosis. The presence of concordant BM involvement can be used to predict relapse and death prior to the calculation of the IPI.
The IPI was originally defined and published in 1993 based upon mathematical modelling of candidate prognostic factors in 2031 patients with de novo diffuse and aggressive lymphoma.4 A revised IPI, which divided DLBCL patients treated with R-CHOP into very good (IPI score=0), good (IPI score=1, 2), and poor (IPI score=3, 4, 5) risk groups reflected the clinical prognosis more distinctively.9 However, in our current study, two groups with a low IPI (IPI score=0, 1, 2) and high IPI (IPI score=3, 4, 5) could be distinguished based on Kaplan–Meier analysis and the log-rank sum test (figure 3). Although a high IPI score was associated with the concordant group, multivariate analysis revealed that it was a negative prognostic factor for PFS and OS independent of the IPI score.
Chung et al3 carried out a similar study to that of our current report prior to the introduction of rituximab, and also found that concordant BM involvement was a negative prognostic factor for lymphoma. Similar results were also obtained by Sehn et al10 who found that concordant BM involvement was a negative prognostic factor independent of the IPI score in patients treated with R-CHOP. In our current study, the data of all three categories (concordant, discordant and negative involvement) were analysed together using the Cox proportional hazard regression model, and not in a pairwise manner.
Our present study differs from the two previous studies mentioned above insofar as the patients we enrolled comprise a relatively homogenous group from a single institute with consistent treatment modalities. This may introduce patient bias. Nonetheless, the number of patients involved in this study represents approximately 9.3% of DLBCL patients in South Korea.11 ,12 The poor prognosis related to the concordant pattern of BM involvement in DLBCL patients was also found in the Asian, mostly Korean, population. Our present study also showed that extensive BM involvement, represented by leukaemic manifestations, was more frequent in the concordant group than in the discordant group. Another study3 has reported that extensive (≥15%) BM involvement is more often seen in cases of concordant rather than discordant BM involvement.
Our current analysis further revealed that discordant involvement was more frequent in clot sections than in trephine biopsies. Nonetheless, we cannot infer any clinical or technical importance to this finding, other than that clot sections helped to detect the involvement of discordant BM involvement. Concordant involvement was more frequently found in both clot sections and trephine biopsies. Analysis of clot sections is useful because it reduces the number of false-negative diagnoses.13 The possibility of false-positive clot sections in our study can be ruled out, given that all findings were confirmed by immunohistochemistry.
In summary, we confirm here, that concordant BM involvement is a negative prognostic factor for PFS and OS in DLBCL patients, independent of IPI factors. However, discordant involvement had no impact on survival in these cases, and had a prognosis similar to that of negative BM involvement. Based on our results, DLBCL patients with concordant BM involvement at diagnosis may be expected to have a greater likelihood of relapse and early death despite treatment. This study is based on a large, relatively homogenous group of patients attending a single institute. Although the nature of concordant and discordant morphology should be investigated in future studies, incorporation of cytomorphology that indicates BM involvement into prognostic evaluations may help to improve the management of these patients.
Take-home messages
-
Multivariate analysis for progression-free survival and overall survival revealed that concordant bone marrow (BM) involvement was a negative prognostic factor, whereas, discordant BM involvement was almost similar to negative involvement.
-
When assessing BM involvement in diffuse, large, B-cell lymphoma for prognostication, not only the presence of lymphoma in BM is important, but also its morphologic classification.
References
Footnotes
-
Contributors All authors contributed equally to this article.
-
Competing interests None.
-
Ethics approval Forum for Ethical Review Committees in Asia and the Western Pacific (FERCAP).
-
Provenance and peer review Not commissioned; externally peer reviewed.