Volume 94, Issue 8 p. 2151-2159
Original Article
Free Access

Correlation between MIB-1 and other proliferation markers

Clinical implications of the MIB-1 cutoff value

Frédérique Spyratos Ph.D.

Corresponding Author

Frédérique Spyratos Ph.D.

Laboratoire d'Oncobiologie, Centre René Huguenin, Saint-Cloud, France

Fax: 33-1-47-11-15-68

Laboratoire d'Oncobiologie, Centre René Huguenin, 35 rue Dailly, 92210 Saint-Cloud, France===Search for more papers by this author
Magali Ferrero-Poüs Ph.D.

Magali Ferrero-Poüs Ph.D.

Laboratoire d'Oncobiologie, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
Martine Trassard M.D.

Martine Trassard M.D.

Département d'Anatomie et Cytologie Pathologiques, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
Kamel Hacène D.Sc.

Kamel Hacène D.Sc.

Département de Statistiques Médicales, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
Edelmira Phillips M.T.

Edelmira Phillips M.T.

Département de Biologie, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
Michèle Tubiana-Hulin M.D.

Michèle Tubiana-Hulin M.D.

Département de Médecine, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
Viviane Le Doussal M.D.

Viviane Le Doussal M.D.

Département d'Anatomie et Cytologie Pathologiques, Centre René Huguenin, Saint-Cloud, France

Search for more papers by this author
First published: 15 April 2002
Citations: 141

Abstract

BACKGROUND

Cell proliferation is a major determinant of the biologic behavior of breast carcinoma. MIB-1 monoclonal antibody is a promising tool for determining cell proliferation on routine histologic material. The objectives of this study were to compare MIB-1 evaluation to other methods of measuring cell proliferation, with a view to refining the cutoff used to classify tumors with low and high proliferation rates in therapeutic trials.

METHODS

One hundred eighty-five invasive breast carcinomas were evaluated for cell proliferation by determining monoclonal antibody MIB-1 staining, histologic parameters (Scarff–Bloom–Richardson grade and mitotic index) on paraffin sections, S-phase fraction (SPF) by flow cytometry, and thymidine-kinase (TK) content of frozen samples.

RESULTS

There was a high correlation (P = 0.0001) between the percentage of MIB-1 positive tumor cells and SPF, TK, histologic grade, and the mitotic index. Multivariate analyses including MIB-1 at 5 different cutoffs (10%, 15%, 17% [median], 20%, 25%) and the other proliferative markers showed that the optimal MIB-1 cutoff was 25% and that the mitotic index was the proliferative variable that best discriminated between low and high MIB-1 samples. A MIB-1 cutoff of 25% adequately identified highly proliferative tumors. Conversely, with a MIB-1 cutoff of 10%, few tumors with low proliferation were misclassified.

CONCLUSIONS

The choice of MIB-1 cutoff depends on the following clinical objective: if MIB-1 is used to exclude patients with slowly proliferating tumors from chemotherapeutic protocols, a cutoff of 10% will help to avoid overtreatment. In contrast, if MIB-1 is used to identify patients sensitive to chemotherapy protocols, it is preferable to set the cutoff at 25%. The MIB-1 index should be combined with some other routinely used proliferative markers, such as the mitotic index. Cancer 2002;94:2151–9. © 2002 American Cancer Society.

DOI 10.1002/cncr.10458

Cell division kinetics is an important predictor of the clinical outcome of breast carcinoma patients.1 Cellular proliferation can be measured using a variety of methods. Mitotic indices have been widely used as part of various tumor grading methods.2-4 DNA synthesis initially was measured in terms of tritiated thymidine incorporation5; proliferation subsequently was measured by using static methods such as flow cytometric determination of the S-phase fraction (SPF).6 Cell proliferation also can be measured by specifically assaying enzymes involved in DNA synthesis, such as thymidine-kinase (TK), an enzyme involved in the one-step salvage pathway of pyrimidine synthesis.7 Immunohistochemical determination of proliferation indices is an expanding area of research, based on detection of antigens present during cell proliferation;8, 9 the widely used Ki-67 now has been supplanted by MIB-1 antibody, which has similar epitope selectivity10 but recognizes the target in paraffin embedded tissues.11 There is no consensus on the best proliferation index, or on the optimal methodology, reagents, or data interpretation. Flow cytometry is widely used, but MIB-1 assays are increasingly popular because of their minimal tissue requirements and suitability to routinely fixed tissues. Results for SPF and MIB-1 are described as “encouraging” in recently published recommendations but not yet adequately characterized for use in clinical practice; in particular, further data from clinical trials based on immunohistochemical (IHC) methods are required.12, 13

The objectives of this study were 1) to technically validate MIB-1 staining before its routine use in our institution, and 2) to compare MIB-1 analysis to other measures of cell proliferation, with a view to refining the cutoff used in therapeutic trials to classify tumors with low and high proliferation rates. The other measures of cell proliferation considered here were the standard and modified Scarff–Bloom–Richardson (SBR and MSBR) histoprognostic grading systems; the mitotic index (included in SBR and MSBR); S-phase measurement by flow cytometry; and TK assay; both the latter used frozen tissue.

MATERIALS AND METHODS

The characteristics of the patients are described in Table 1. From September 1998 to July 2000, 185 patients with invasive breast carcinomas of sufficient size (macroscopic specimens) were enrolled in this study. Immediately after confirmation of invasive carcinoma on frozen sections, part of the tumor was frozen and kept in liquid nitrogen for flow cytometry and TK assay. All tumor specimens were embedded in paraffin and stained with hematoxylin and eosin. After histologic analysis, 159 tumors were classified as invasive ductal carcinomas, 15 as invasive lobular carcinomas, and 11 as other types. All the invasive carcinomas apart from six mucinous carcinomas were graded using the SBR2 and MSBR methods,3 based on the sum of scores for two parameters of nuclear grading (mitotic index and nuclear pleomorphism). Mitotic activity was quantified by microscopic examination of 10 high-power fields. Lymph node status (pN) was determined on axillary lymphadenectomy specimens. Median age at diagnosis was 60 years (range, 33–99 years). Median tumor size was 22 mm (range, 10–53 mm).

Table 1. Characteristics of the Patients and Correlation between MIB-1 Values and Clinical, Histologic, and Biologic Variables
Variable n (%) MIB-1
Mean (SD) Median (range) P value
Age (yrs) 0.689
 ≤ 50 46 (25) 22.50 (20.33) 20 (1–90)
 > 50 139 (75) 20.91 (19.04) 15 (1–80)
Tumor size (pT) (mm) 0.112
 ≤ 20 86 (47) 19.19 (18.93) 12.5 (1–78)
 > 20 98 (53) 22.87 (19.49) 20 (1–90)
Lymph node status (pN) 0.636
 pN = 0 111 (64) 21.36 (20.43) 15 (1–90)
 1 ≤ pN ≤ 3 46 (26) 22.35 (17.37) 20.5 (1–80)
 pN > 3 18 (10) 22.44 (21.16) 17.5 (1–75)
SBR grade 0.0001
 I 30 (17) 7.13 (6.97) 5 (1–25)
 II 90 (50) 19.44 (17.50) 15 (1–75)
 III 59 (33) 32.10 (21.04) 30 (1–90)
MSBR 0.0001
 1 87 (49) 11.62 (13.91) 8 (1–70)
 2 92 (51) 30.95 (19.39) 30 (1–90)
Mitotic index 0.0001
 1 39 (22) 6.51 (6.50) 5 (1–25)
 2 58 (32) 16.05 (16.73) 10 (1–70)
 3 82 (46) 32.60 (19.06) 30 (1–90)
Histologic type 0.552
 Invasive ductal 159 (86) 22.19 (20.16) 20 (1–90)
 Invasive lobular 15 (8) 16.93 (13.75) 15 (1–50)
 Others 11 (6) 14.36 (9.74) 15 (1–35)
Estrogen receptor status 0.0004
 Negative 34 (18) 34.74 (25.47) 27.5 (1–90)
 Positive 150 (82) 18.39 (16.29) 15 (1–75)
Progesterone receptor status 0.0009
 Negative 56 (30) 29.62 (23.43) 24.5 (1–90)
 Positive 128 (70) 17.82 (16.05) 13 (1–75)
DNA ploidy index 0.0001
 Diploid 74 (40) 13.65 (15.01) 10 (1–75)
 Aneuploid 111 (60) 26.41 (20.24) 25 (1–90)
S-phase fraction 0.0001
 Low 53 (34) 13.47 (14.67) 5 (1–60)
 Intermediate 51 (32) 16.51 (12.04) 15 (1–50)
 High 54 (34) 31.85 (23.10) 29.5 (1–80)
Thymidine kinase 0.0001
 Low 59 (34) 12.27 (10.71) 10 (1–45)
 Intermediate 59 (34) 22.80 (17.56) 20 (1–80)
 High 57 (32) 29.54 (24.37) 30 (1–90)
  • SD: standard deviation; SBR: Scarff–Bloom–Richardson; MSBR: modified SBR.

Immunohistochemistry

Tumor samples were fixed for 12–24 hours in ready-to-use AFA (Carlo Erba reagenti, Rodano, Italy), containing absolute ethanol (75% v/v), formol 40% (2% v/v)/acetic acid (5% v/v)/water (18% v/v) and then routinely embedded in paraffin blocks. Immunoperoxidase staining for MIB-1 was performed with an automated avidin-biotin complex method in a Techmat 500 device (Dako SA, Trappes, France). The IHC assay was performed on 4-μm sections cut from the blocks, float-mounted on adhesive (silanized) glass slides (Tespa-coated slides), and left at 50 °C overnight on a slide warmer. The key features of the IHC assay included antigen retrieval in boiling 0.1 M citrate buffer (pH 6.0) in a pressure cooker (4 minutes); blocking of endogenous peroxidase activity with 1% hydrogen peroxide for 15 minutes; incubation with mouse monoclonal antibody MIB-1 (Immunotech, Marseille, France) at a dilution of 1:100 for 1 hour; linking with a rabbit biotinylated antibody against mouse immunoglobulin G at a dilution of 1:100 for 30 minutes; enzyme labeling with streptavidin-horseradish peroxidase (Dako) at a dilution of 1:100 for 30 minutes; chromogen development with 0.03% hydrogen peroxide and 1 mg/ml diaminobenzidine; and counterstaining with Mayer's hematoxylin (1 minute in a 1:5 dilution), without differentiation in acid alcohol. Appropriate positive and negative controls were run with each batch.

MIB-1 Index

The MIB-1 index was determined by using a semiquantitative visual method. Tumor cells were considered positive for MIB-1 only when clear nuclear staining was seen. The percentage of positive neoplastic nuclei was estimated after scanning the entire tumor surface at low power (×10 objective), including areas of highest and lowest positivity. After this first analysis, a quantitative assessment was made at ×20 and/or ×40 magnification by counting a total of 200–500 tumor cells within representative fields. All nuclei with homogeneous granular staining, multiple speckled staining or nucleolar staining were counted as positive, regardless of staining intensity. Cells with cytoplasmic staining were excluded. This evaluation was performed independently by two observers (V.L.D. and M.T.), and an interobserver reproducibility test was performed on the entire series (n = 185).

Estrogen Receptor and Progesterone Receptor Assays

The hormone receptor status of the tumor was recorded at the time of surgery. The tumor was considered to be steroid receptor positive if estrogen receptor (ER) and progesterone receptor (PR) values exceeded 15 fmol/mg protein by enzyme immunoassay (Abbott Laboratories, North Chicago, IL).

Flow Cytometric DNA Analysis and S-Phase Measurement

Cell preparation was standardized as follows: 1) quick specimen thawing; 2) tumor imprints performed systematically, stained with May-Grunwald-Giemsa and observed by a pathologist to check for the presence of malignant cells; 3) mechanical dissociation in phosphate buffer saline; and 4) DNA staining according to Vindelov's method. Flow cytometry was performed on a FACScalibur device (Becton Dickinson, Mountain View, CA). The cell cycle analysis was performed with the Modfit LT 2.0 program (Verity Software House, Topsham, ME). The DNA-diploid peak was located on DNA histograms by using an external standardization procedure with normal human lymphocytes positioned at the fifth part of the red fluorescence scale. DNA ploidy and the SPF were obtained after gating on a dot plot (signal peak width vs. signal peak area), selecting a representative amount of debris and excluding doublets.

Rules established during a previous interlaboratory control procedure14 were applied when using the different cell cycle software models and for objective interpretation of DNA histograms. They included graphic aggregate subtraction. The limits for the ploidy index were hypodiploid less than 0.95; 0.95 less than or equal to diploid less than or equal to 1.1; aneuploid greater than 1.10; 1.90 less than or equal to tetraploid less than or equal to 2.05. Multiploid tumors were characterized by DNA histograms showing two or more aneuploid G0G1 peaks. Automatic procedures were not used. We calculated SPF when the coefficient of variation of the G0G1 peaks was lower than 5%, debris represented less than 20% of all acquired events, and a minimal aneuploid fraction was present. When a unimodal histogram was obtained, the diploid option of the software programs was used. If DNA content was abnormal, only the aneuploid SPF was taken into account (the diploid SPF was not calculated). In every case, the rectangular option was chosen for SPF calculation. The debris subtraction option was always used. SPF adjusted for ploidy was categorized as low, intermediate, or high according to the 33rd and 66th percentiles.

TK Assay

Cytosols prepared for hormone receptor assays at the time of initial surgery were stored in liquid nitrogen until analysis. The cytosolic protein concentration was determined using the BCA assay (Pierce, Rockford, IL) with the same standard for all samples.

Thymidine-kinase enzyme activity was determined using the Prolifigen TK-REA assay (Sangtec Medical, Bromma, Sweden), which is optimized for the TK1 isoenzyme. Measurements were performed as previously described15 with the following modifications recommended by the EORTC Receptor and Biomarker Study Group16: cytosols were diluted 1:20 in heat-inactivated normal calf serum containing 90 mM Hepes, 18 mM KCl, 6 mM dithiotreitol, 8 mM MgCl2, and 4 mM ATP; TK levels were expressed in milliunits per milligrams of protein. Thymidine-kinase values were categorized as low, intermediate, or high according to the 33rd and 66th percentiles.

Statistical Methods

Statistical analysis was performed using the SAS statistical package (SAS Institute, Cary NC).

Descriptive statistics were calculated for all proliferation markers. Correlations between MIB-1, SPF, and TK were identified with Spearman rank coefficient. Associations with other clinical or pathologic variables were sought using Mann–Whitney or Kruskal–Wallis tests.

The Cox logistic regression model was used to determine which variables best discriminated samples with low MIB-1 from those with high MIB-1, as defined by five different cutoff values. Each model was obtained by a stepwise selection procedure and was evaluated using two statistics, namely, the Harrel c-index, i.e., the area under the receiver operating characteristics curve, because the response is binary (the higher the statistic, the better the model); and the Akaike information criterion (AIC) to compare the five models, that with the lowest value being the preferred model.

RESULTS

Correlations between MIB-1 values and other clinical and pathologic variables are shown in Table 1. High MIB-1 was associated with SBR histologic Grade III (P = 0.0001), high MSBR grade (P = 0.0001), a high mitotic index (P = 0.0001), ER negativity (P = 0.0004), and PR negativity (P = 0.0009). MIB-1 did not correlate with age, tumor size, lymph node status, or histologic type. The distribution of MIB-1, TK and SPF is shown in Table 2. The overall median MIB-1 value was 17%. Calculation of SPF was possible for 158 (85%) of the 185 tumors submitted for flow cytometry. The median SPF was 3.29%. TK was determined in 175 tumors (95%), and the median value was 90.18 mU/mg protein.

Table 2. Distribution of MIB-1, SPF, and TK Values
Variable n Mean (SD) Median Range
MIB-1 (%) 185 21.30 (19.33) 17 1–90
SPF (%) 158 4.96 (4.62) 3.29 0.19–26.10
TK (mU/mg protein) 175 453.06 (2383) 90.18 0–30581
  • SPF: S-phase fraction; SD: standard deviation; TK: thymidine-kinase.

Association between MIB-1 and Other Proliferative Factors

High MIB-1 was associated with a high mitotic index (P = 0.0001). Median MIB-1 values were significantly higher in DNA aneuploid tumors than in diploid tumors. There was a close relation between MIB-1 values and SPF (Spearman ρ = 0.52; P = 0.0001). Regarding ploidy status, the correlation between MIB-1 and SPF was detected only in aneuploid tumors (ρ= 0.57; P = 0.0001; ρ value for diploid tumors = 0.22; P = 0.066). There was a significant relation between MIB-1 values and TK values (Spearman ρ = 0.33; P = 0.0001).

The mean and median SPF values increased proportionally to the MIB-1 score (Table 3), and there was no major variability among tumors with the same MIB-1 scores.

Table 3. Quantitative Correlation between MIB-1 and SPF
MIB-1 (%) n SPF (%)
Mean SD Median
≤ 10 66 2.88 2.81 2.10
11–20 27 3.48 1.81 3.64
21–30 27 5.55 3.86 6.05
31–40 17 7.06 4.40 7.78
41–50 10 10.67 5.00 9.65
> 50 11 11.16 8.11 8.51
Total 158 4.96 4.62 3.29
  • SPF: S-phase fraction; SD: standard deviation.

When TK values were used instead of SPF values (Table 4), these comparisons remained significant but failed to show a monotonous increase in the mean and median TK values.

Table 4. Quantitative Correlation between MIB-1 and TK
MIB-1 (%) n TK (mU/mg protein)
Mean SD Median
≤ 10 72 296.45 898.39 64.49
11–20 28 112.01 138.26 73.30
21–30 32 197.66 348.04 99.72
31–40 20 301.46 337.66 196.76
41–50 9 3815.48 10063.23 184.68
> 50 14 579.40 577.03 388.22
Total 175 453.06 2382.61 90.18
  • TK: thymidine-kinase; SD: standard deviation.

MIB-1 Categorization and Cox Logistic Regression Analysis

Five cutoffs (10%, 15%, 17% [median], 20%, and 25%) were used to dichotomize MIB-1 values into those reflecting low and high proliferation rates.

Each MIB-1 cutoff value then was entered as a dependent variable into a logistic regression model, to determine which of the variables listed in Table 1 best discriminated high from low MIB-1 samples. Each model was obtained by a stepwise selection procedure and was evaluated with multiple statistics (Table 5). Whatever the cutoff used for MIB-1, the mitotic index was always the most discriminant variable, followed by SPF when the MIB-1 cutoff was set at 10%, or by TK and the ploidy index when the MIB-1 cutoff was set at 25%.

Table 5. Results of the Cox Logistic Regression Model
MIB-1 cutoff (%) Discriminant variables P value Harrel's c-indexa (ROC AUC) AIC
10 Mitotic index 0.0001 0.838 136.467
SPF 0.025
15 Mitotic index 0.0001 0.823 138.677
17 Mitotic index 0.0001 0.809 142.592
20 Mitotic index 0.0001 0.802 144.895
25b Mitotic index 0.0001 0.850 129.692
TK 0.0033
Ploidy index 0.0329
  • ROC: receiver operating characteristic; AUC: area under the curve; AIC: Akaike information criterion.
  • a Harrel's c-index was highest in the model in which the MIB-1 cutoff was set at 25% (the higher the statistic, the better the model).
  • b The model with the cutoff set at 25% had the lowest AIC statistic and thus was the preferred model.

The Harrel c-index was highest in the model in which the MIB-1 cutoff was set at 25% (the higher the statistic, the better the model). With the AIC statistic, the model with the cutoff set at 25% had the lowest value and thus was the preferred model.

Distribution of Other Proliferation Markers According to the MIB-1 Cutoff

The distribution of the other proliferation markers is shown in Table 6 according to the five MIB-1 cutoffs. Among the tumors with MIB-1 staining below 10%, 19% (11 of 59) of the overall population and 17% (4 of 24) of the aneuploid tumors were classified as high SPF; 16% (10 of 62) were MSBR Grade 2 and 11% (7 of 62) had a mitotic index of 3. Among the tumors with MIB-1 staining above 10%, 78% (77 of 99) of the overall population and 81% (51 of 63) of the aneuploid tumors were classified as intermediate or high SPF; 70% (82 of 117) were MSBR Grade 2 and 91% (107 of 117) had a mitotic index of 2 or 3.

Table 6. Distribution of Other Proliferation Markers According to the MIB-1 Cutoff
Variable n MIB-1 ≤ 10%, n (%) MIB-1 ≤ 15%, n (%) MIB-1 ≤ 17%,a n (%) MIB-1 ≤ 20%, n (%) MIB-1 ≤ 25%, n (%) MIB-1 > 25%, n (%)
SPF
 Low 53 31 (52) 33 (48) 36 (44) 36 (44) 40 (40) 13 (23)
 Intermediate 51 17 (29) 23 (33) 29 (36) 29 (35) 38 (38) 13 (23)
 High 54 11 (19) 13 (19) 16 (20) 17 (21) 23 (22) 31 (54)
Aneuploid SPF
 Low 28 16 (66) 16 (62) 17 (57) 17 (57) 20 (48) 8 (18)
 Intermediate 28 4 (17) 6 (23) 9 (30) 9 (30) 16 (38) 12 (27)
 High 31 4 (17) 4 (15) 4 (13) 4 (13) 6 (14) 25 (55)
MSBR
 1 87 52 (84) 62 (80) 67 (75) 67 (74) 74 (67) 13 (19)
 2 92 10 (16) 16 (20) 22 (25) 23 (26) 36 (33) 56 (81)
Mitotic index
 1 39 29 (47) 33 (42) 35 (39) 35 (39) 38 (35) 1 (1)
 2 58 26 (42) 35 (45) 38 (43) 38 (42) 43 (39) 15 (22)
 3 82 7 (11) 10 (13) 16 (18) 17 (19) 29 (26) 53 (77)
  • SPF: S-phase factor; MSBR: modified Scarff–Bloom–Richardson.
  • a 17% is the median value in our series.

Among the tumors with MIB-1 staining below 25%, 23% (22 of 101) of the overall population and 14% (6 of 42) of the aneuploid tumors were classified as high SPF; 33% (36 of 110) were MSBR Grade 2, and 26% (29 of 110) had a mitotic index of 3. Among the tumors with MIB-1 staining above 25%, 77% (44 of 57) of the overall population and 82% (37 of 45) of the aneuploid tumors were classified as high SPF; 81% (56 of 96) were MSBR Grade 2 and 99% (68 of 69) had a mitotic index of 2 or 3.

Intermediate cutoffs (15%, 17%, or 20%) were less discriminatory.

Most of the tumors considered to have low proliferation when the MIB-1 cutoff was set at 10% also were considered to have low proliferation with the other methods. When the MIB-1 cutoff was set at 25%, a significant number of tumors considered to have low proliferation had high proliferation with the other methods. Conversely, most tumors with MIB-1 values above 25% also were considered to have high proliferation with the other methods.

DISCUSSION

Several approaches have been developed to study tumor proliferation, including a labeling index,5 SPF measurement by flow cytometry,6, 17-19 immunohistochemical detection of proliferation-associated antigens,20-29 and cytosolic TK assay.7, 15 Highly proliferative breast tumors are associated with shorter patient survival, whatever the method used to measure proliferation.5, 7, 18, 20, 22, 24 A strong correlation generally is observed between these methods, even though they do not measure the same biologic entities. Studies mainly based on flow cytometric DNA analysis have suggested that highly proliferative tumors show increased sensitivity to neoadjuvant30, 31 and adjuvant chemotherapy,17, 19 regardless of the impact of such treatment on patient survival. In contrast, the rationale of chemotherapy for slowly proliferating tumors is controversial.6, 7, 14 Hence, proliferative markers have the potential to distinguish patients with rapidly proliferating tumors that are likely to respond to chemotherapy from patients with slowly proliferating tumors who may not need aggressive treatment. However, if they are to be used in clinical trials, these factors must be determined with techniques that are accurate, reproducible, and available in most laboratories. Despite the many studies focusing on proliferative markers in breast carcinoma, the American Society of Clinical Oncology tumor markers expert panel considered that data on SPF and immunohistochemical indices are insufficient to recommend their routine use.13 However, in practice, monoclonal antibody MIB-1 is probably the most widely used proliferative marker. It reacts with an antigen that is only present in the nucleus of proliferating cells, has similar epitope sensitivity to Ki-67, and, contrary to the latter, can be used with paraffin sections.

In this study, we technically validated the use of MIB-1 in breast carcinoma and assessed the clinical implications of the choice of MIB-1 cutoff. MIB-1 scores were compared with the results of flow cytometric SPF determination and TK assay. MIB-1 scores also were compared with other classic factors measuring the proliferation rate, namely, SBR and MSBR grades and the mitotic index, which is a component of histologic grading system. We used five different MIB-1 cutoffs, including the median value in our series and values chosen in previous series, to define the most appropriate cutoff for distinguishing between tumors with low and high proliferation rates.

The distribution of MIB-1 values in this series (mean and median) were consistent with those found in previous studies (Table 7). Reported links between MIB-1 and other known prognostic factors also were confirmed here. MIB-1 values tended to be higher in patients younger than 50 years than in older patients, and the lack of statistical significance was probably because there were few patients younger than 50 years in our series. Some authors have found a similar correlation,20, 27, 28 whereas others have not.22, 24, 26 We observed no correlation between MIB-1 and tumor size, as previously reported.22, 26 When such a correlation is observed, it is usually weak.21, 24, 27, 28 As generally reported,21, 22, 28 MIB-1 values were lower in lobular carcinomas than in ductal carcinomas, but the difference was not significant in our series, probably because there were few lobular carcinomas. The lack of association with lymph node status in our study is consistent with previous reports.25, 26 MIB-1 correlated negatively with hormone receptor status, in keeping with most other studies.24-28

Table 7. Distribution of MIB-1 Values in Studies of Paraffin Sections of Breast Tumors
Authors No. of patients (type of tumors) Antibody used MIB-1 distribution Cutoffs used for MIB-1
Arber et al. (1997)32 48 (T1–T2, N0) MIB-1 Median, 16.4% in patients without recurrence Median
Median, 36.7% in patients with recurrence
Clahsen et al. (1999)20 441 (pN0) MIB-1 NS 20%
Dettmar et al. (1997)22 90 (pN0) MIB-1 Median, NS 25%
Mean, 18%
Range, 1–81%
Ellis et al. (1996)23 75 (postmenopausal) MIB-1 Median, 9% ns
Mean, NS
Range, 1–83.4%
Jansen et al. (1998)24 341 MIB-1 Median, 7% 7%
Mean, NS
Range, 0.71–11%
Keshgegian and Craan (1995)25 135 MIB-1 Median, 8.2% 10%
Mean, NS
Range, 0–50%
MacGrogan et al. (1997)26 112 MIB-1 Median, 27.5% Median
Mean, NS
Range, 1–95%
Rudolph et al. (1999)27 371 (pN0) Ki-S5 (=MIB-1) Median, 23.5% 25%
Mean, NS
Range, 3–93%
Thor et al. (1999)28 486 MIB-1 Median, 28.6% Median
Mean, 32.2%
Range, 0–99%
  • NS: not specified.

MIB-1 correlated with histologic tumor grade. Other authors also have observed a significant relation between MIB-1 and tumor grade in breast carcinoma.21, 22, 25-29, 32 MIB-1 also correlated strongly with MSBR, the modified SBR grade defined by the mitotic index and nuclear pleomorphism. Finally, we confirmed that the mitotic index, a component of the SBR grading system, correlates strongly with MIB-1.20, 25, 26, 28, 29, 32

Regarding links between MIB-1 and the other proliferation markers, we confirmed the good correlation between MIB-1 and SPF measured on frozen tumors.23, 25, 26 We also confirmed that the overall correlation between MIB-1 and SPF was primarily because of aneuploid tumors.22 This may be explained by the finding that flow cytometric DNA analysis tends to underestimate SPF in diploid tumors.

We also observed a correlation between TK and MIB-1, a relation that, to our knowledge, has not been examined previously. A nonlinear relation was observed between TK values and MIB-1, indicating that TK activity represents a different aspect of proliferation, even though the clinical significance of TK was close to that of SPF in a recent large multicenter study.7

Thus, we confirm that MIB-1 is strongly related to histologic and nuclear grade, the mitotic index, and SPF (mainly in aneuploid tumors) and further show that MIB-1 is related to TK, albeit to a lesser degree.

To refine the correlation studies, we used more sophisticated statistical models. Logistic regression analysis showed that the mitotic index was the proliferation variable that best discriminated high from low MIB-1 samples. This is consistent with the finding that KI-67 antigen expression increases during progression of the mitotic cycle, increasing during the latter half the SPF and peaking in the G2 and M phases.9 The mitotic index is a rapid and cost-effective tool for estimating tumor cell proliferation, and reasonable reproducibility can be achieved with a strictly standardized methodology.4 In clinical trials, MIB-1 could be used in conjunction with the mitotic index to ensure correct tumor classification on the basis of proliferative potential.

Regarding the choice of MIB-1 cutoff, several statistical tests applied to the logistic regression analyses indicated that a cutoff of 25% was optimal. This value is within the range (7– 36.5%) of values in most published series of breast carcinoma, as shown in Table 7. The median often is used as a cutoff, with no rational justification, and there is no consensus on the best cutoff for MIB-1, possibly owing to the lack of a published quality insurance program comparable to those used for HER-2 and hormone receptors.33, 34 Another possible explanation for the wide range of MIB-1 cutoffs is the clinical heterogeneity of published series. To our knowledge, no attempt has been made to use other proliferation markers to check the accuracy of MIB-1–based tumor categorization. We found that the choice of MIB-1 cutoff influenced the accuracy of this marker, as judged on the basis of MSBR grade, the mitotic index, SPF, and TK. With a MIB-1 cutoff of 10%, few tumors with low proliferation were misclassified. Conversely, a MIB-1 cutoff of 25% acceptably identified highly proliferative tumors. There is no obvious difference between the cutoff set at 10% or 25% regarding comparison with SPF. The difference is more pronounced when MIB-1 is compared with MSBR (16% vs. 33%) or the mitotic index (11% vs. 26%). This is not surprising, because the associations between MIB-1 and MSBR and between MIB-1 and the mitotic index are stronger than that between MIB-1 and SPF. Hence, the choice of MIB-1 cutoff will depend on the following clinical objective: if MIB-1 is used to exclude patients with slowly proliferating tumors from chemotherapy protocols, a cutoff of 10% will help to avoid overtreatment. In contrast, if MIB-1 is used to identify tumors sensitive to chemotherapy, it seems preferable to set the cutoff at 25%. At all events, MIB-1 should be combined with some other routinely used proliferative markers, such as the mitotic index.