The MIB-1 labeling index, which is based on Ki67 immunostaining, is widely used to evaluate the proliferation of tumor cells in lymphoma. However, its clinical significance has not been fully assessed. We retrospectively evaluated the prognostic impact of the MIB-1 labeling index at the time of diagnosis, in 98 patients with follicular lymphoma (FL) grade 1–3b who were treated uniformly with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisolone (R-CHOP) therapy. The 5-year progression-free survival (PFS) for an MIB-1 labeling index of ≥10% (n = 60) and <10% (n = 38) was 35% and 61%, respectively (P = 0.015). The 5-year overall survival (OS) for an MIB-1 labeling index of ≥10% and <10% was 77% and 92%, respectively (P = 0.025). Pathological grading was not correlated with PFS or OS. In multivariate analysis, an MIB-1 labeling index of ≥10% was independently associated with poor PFS and OS. In conclusion, an MIB-1 labeling index of 10% is a useful cut-off level for predicting the prognosis of patients with FL.
Follicular lymphoma (FL) is an indolent non-Hodgkin's lymphoma, with a highly variable clinical course that can range from indolent to rapidly progressive disease, including the transformation to aggressive non-Hodgkin's lymphoma. The classification system of the World Health Organization (WHO) is widely used to grade FL; this grading is based on the number of centroblasts that can be counted in representative malignant follicles under a microscope with a standard 40× high power field. When FL patients are given the same treatment, irrespective of the disease grade, patients with grade 3 FL have a worse prognosis than those with grade 1 or 2 FL. However, this difference disappears when the intensity of treatment for patients with grade 3 FL is increased by the addition of anthracycline-based regimens.[2, 3]
The MIB-1 labeling index, which is based on Ki67 immunostaining, is widely used for evaluating the proliferation of tumor cells, because the mAb Ki67 reacts with an undefined antigen present in the nuclei of cells in all cell cycle phases, except G0. The MIB-1 labeling index can be used as a prognostic factor in patients with FL and other aggressive lymphomas. However, the prognostic impact of Ki67 was changed after the introduction of rituximab therapy. Follicular lymphoma is not a curative disease even if patients are treated with rituximab; therefore, it is crucial to identify FL patients who require aggressive therapy because they have a high risk of developing progressive or transformed disease early.
The purpose of this study was to evaluate the prognostic impact of the MIB-1 labeling index, calculated at the time of diagnosis, in FL patients who were treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisolone (R-CHOP) alone.
Materials and Methods
Between 2001 and 2009, we enrolled 98 consecutive FL patients diagnosed at one of the seven participating hospitals. Since 2001, the Yokohama City University Hematology Group in Japan has uniformly treated FL patients with six cycles of standard R-CHOP therapy for 21 days, except for those with stage 1 disease. Patients who had a partial response after the first four R-CHOP treatment cycles were given a total of eight treatment cycles, and patients who did not achieve partial response after the first four cycles, or those who showed disease progression at any given time, received salvage therapy. Patients who had a bulky mass, defined as any mass of ≥10 cm in diameter in a horizontal plane radiograph or a mediastinal mass with a maximum diameter exceeding one-third of the maximum chest diameter, received irradiation following complete response with six to eight cycles of R-CHOP. Patients who required a dose reduction of more than 20% were excluded from this study.
Pathological diagnosis was made by the pathologists at each of the participating hospitals; the biopsied specimens (paraffin-embedded tissue sections) were reviewed according to the WHO classification. Thereafter, the diagnosis and the proliferation potential of lymphoma cells were centrally reviewed by three hematological pathologists (S.S., N.T., and K.T.) at the Cancer Institute, Japanese Foundation for Cancer Research (Tokyo, Japan). The reviewed diagnosis was used in this study.
In detail, we heated the slides for 45 min at 102°C in antigen retrieval solution (Nichirei Bioscience, Tokyo, Japan). For conventional immunostaining, the slides were incubated at room temperature with primary antibodies anti-CD10 mouse mAb (clone 56C6, dilution 1:100; Leica Biosystems, Newcastle, UK), anti-CD20 mouse mAb (clone L26, dilution 1:50; DakoCytomation, Glostrup, Denmark), anti-BCL2 mouse mAb (clone 124, dilution 1:100; DakoCytomation), and anti-Ki67 mouse mAb (clone MIB-1, dilution 1:50; DakoCytomation) for 30 min. The immune complexes were then detected with polymer reagent (Histofine Simple Stain MAX PO; Nichirei Bioscience). Color was developed with 3,3′- diaminobenzidine tetrahydrochloride (DAB) using the DAB substrate kit (Nichirei Bioscience). The MIB-1 labeling index was measured by Ki67 immunostaining on the serial section of the paraffin-embedded biopsy sample. It was automatically counted in a unit area by computer analysis at the Cancer Institute of the Japanese Foundation for Cancer Research. The count was carried out in the intrafollicular area by selecting five areas of 500-μm diameter per specimen. The median number of counted cells used to determine the MIB-1 labeling index was 13 947 (range, 4766–18 022).
This study was approved by the Yokohama City University Hospital Clinical Research Ethics Board (Yokohama, Japan), and it was carried out in accordance with the Declaration of Helsinki.
Progression-free survival (PFS) was calculated from the date of therapy initiation to the date of progression, and overall survival (OS) was calculated from the date of therapy initiation to the date of final follow-up or death from any cause. Kaplan–Meier survival analysis was carried out to estimate PFS and OS, with the difference between the curves examined by the log–rank test. Receiver operating characteristic (ROC) curve analysis was used to describe the association between the sensitivity and specificity of the MIB-1 labeling index at diagnosis to decide the appropriate cut-off level. The ROC curves and the respective areas under the ROC curves were calculated. A Cox proportional hazards model was used to determine hazard ratios (HRs), and 95% confidence intervals (CIs) were used in univariate and multivariate analysis. Predictors with significance (P < 0.05) and borderline significance (P < 0.1) in the univariate analysis were entered into multivariate analysis. A P-value <0.05 was considered statistically significant. Statistical analysis was carried out with R 2.13.2 statistical software (T Development Core Team, Vienna, Austria).
The characteristics of the patients are summarized in Table 1. The median age was 57 years (range, 34–85 years). According to the WHO pathological grading, grade 1 FL was observed in 63 patients (64%), grade 2 FL in 21 patients (21%), grade 3a FL in 12 patients (12%), and grade 3b FL in 2 patients (2%).
Table 1. Baseline characteristics of patients with follicular lymphoma (FL) and correlations with MIB-1 labeling index on diagnosis (n = 98)
Characteristic or therapeutic factor
No. of patients (%)
MIB-1 labeling index, %
B2M, beta 2 microglobulin; BM, bone marrow; ECOG, Eastern Cooperative Oncology Group; FLIPI, Follicular Lymphoma International Prognostic Index; LDH, lactate dehydrogenase; NT, not tested.
ECOG performance status
At least 3
MIB-1 labeling index, %
MIB-1 labeling index
The median MIB-1 labeling index was 12.9% (range, 0.8–79.9%). Based on the results of the ROC curve, an MIB-1 labeling index of 10% was chosen as the cut-off level. The ROC curve is shown in Figure 1. An MIB-1 labeling index of 9.9% was the most appropriate cut-off level for predicting the PFS (sensitivity, 0.75; specificity, 0.49). However, an MIB-1 labeling index of 13.3% was found to be the best for predicting the OS (sensitivity, 0.82; specificity, 0.58). The areas under the curves were 0.61 in PFS and 0.67 in OS. We used an MIB-1 labeling index of 10%, 13%, and 15% to determine the most suitable cut-off level, and we found significant differences for prognosis at all cut-off levels. For PFS, the P-values were 0.015, 0.048, and 0.04 when an MIB-1 labeling index of 10%, 13%, and 15%, was used as the cut-off level, whereas the P-values were 0.009, 0.002, and 0.006 when an MIB-1 index of 10%, 13%, and 15%, respectively, was used. Therefore, an MIB-1 labeling index of 10% was thought to be the most appropriate cut-off level.
Outcome and prognostic factors
The PFS and OS for the entire cohort of 98 patients are shown in Figure 2. The median observation period for surviving patients was 60 months. The estimated 5-year PFS and OS rates were 48% and 83%, respectively. The 5-year PFS for an MIB-1 labeling index of ≥10% (n = 60) and <10% (n = 38) was 35% and 61%, respectively (P = 0.015). The 5-year OS for an MIB-1 labeling index of ≥10% and <10% was 77% and 92%, respectively (Fig. 3). Pathological grading was not correlated with the PFS (5-year PFS of 49% vs 42%, P = 0.87) or OS (85% vs 69%, P = 0.28) (Fig. 4). The following variables were assessed for their impact on PFS and OS, age, gender, clinical stage, ECOG performance status, FL international prognostic index (FLIPI) scores, presence or absence of “B” symptoms, pathological grade, existence of bulky mass, bone marrow involvement, lactate dehydrogenase level, and the MIB-1 labeling index (Table 2).
Table 2. Prognostic impact of each factor, including MIB-1 labeling index, on progression-free survival (PFS) and overall survival (OS) in patients with follicular lymphoma (n = 98), univariate and multivariate analyses
Characteristic or therapeutic factor
5-year survival, %
5-year survival, %
HR (95% CI)
HR (95% CI)
Predictors with borderline significance (P < 0.1) in univariate analysis were entered into multivariate analysis. BM, bone marrow; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; FLIPI, Follicular Lymphoma International Prognostic Index; HR, hazard ratio; LDH, lactate dehydrogenase; NS, not significant; NT, not tested.
ECOG performance status
MIB-1 labeling index, %
Multivariate Cox proportional hazards analysis of the variables (those with P-values <0.1 in the univariate analysis) revealed that patients with stage IV FL (HR, 2.26; P = 0.019) and an MIB-1 index of ≥10% (HR, 2.61; P = 0.0086) had significantly worse PFS, whereas patients with high FLIPI (HR, 3.64; P = 0.031) and an MIB-1 index of ≥10% (HR, 3.93; P = 0.042) had significantly worse OS.
This was a multicenter retrospective study to evaluate the prognostic impact of the MIB-1 labeling index at the time of diagnosis of FL. Both PFS and OS of 98 patients with FL were significantly correlated with the MIB-1 labeling index of under or over 10%. This finding was also confirmed by the multivariate analysis.
Koster et al. also reported that the MIB-1 labeling index was significantly correlated with PFS (P = 0.0006) and OS (P = 0.002) in a cohort of 64 patients with FL. Furthermore, similar to our findings, they reported that the histological grade did not predict the outcome. Miller et al. and Martin et al. found that the MIB-1 labeling index correlated with OS but not with PFS. However, in the study of Martin et al., significance was not detected by the multivariate analysis. Wang et al. reported that patients with low-grade FL and a high proliferation index determined by MIB-1 labeling index tended to have a shorter OS (P = 0.087) than those with a low proliferation index. In addition, Saito et al. found that FL patients with a high MIB-1 labeling index tended not to achieve complete response after treatment including rituximab. In contrast, Llanos et al. found no correlation between the MIB-1 labeling index and OS in 49 FL patients.
Patients in the studies described here were not given uniform treatment. Therefore, to exclude therapeutic bias, we enrolled patients who were treated with standard R-CHOP therapy. In addition, the cut-off level of the MIB-1 labeling index varied between reports. We suggest that an MIB-1 labeling index of 10% is a useful histological factor that affects the outcome of patients with FL.
In our study, the histological grade of FL did not predict the outcome, PFS, or OS of patients. Koster et al. also reported that in the pre-rituximab era, FL grade did not predict the outcome. A major concern for histological grading is poor reproducibility because of the complications associated with the manual counting of centroblasts.[11, 12] In 41 of the patients in our study, the histological grade diagnosed by the initial pathologist differed from the diagnosis made during the histological review (data not shown). In contrast, the inter-observer variation in evaluating the MIB-1 labeling index would be smaller because of the quantitative evaluation by immunostaining compared with the qualitative method of histological grading.
In conclusion, an MIB-1 labeling index of 10% is a useful cut-off level for predicting the prognosis of FL patients treated with R-CHOP therapy. The multivariate analysis also indicated a significant correlation between prognosis and an MIB-1 labeling index of 10%.