Search

Background The prognostic role of CD68+ tumor-associated macrophages (TAMs) and tumor infiltrating T-cells including FoxP3+ regulatory T-cells (Tregs) has been extensively evaluated in areas of lymphoma research, however their expression and prognostic role have little been explored in primary central nervous system lymphoma (PCNSL). Therefore, we investigated CD68 and FoxP3 expression in tumor microenvironment of PCNSL and evaluated its prognostic role. Methods Seventy-six consecutive immunocompetent patients diagnosed with PCNSL between December 2004 and April 2015 treated homogenously with high-dose methotrexate (HD-MTX)-based chemotherapy as an initial treatment in Severance Hospital, Seoul, S. Korea and for whom archived formalin-fixed and paraffin-embedded (FFPE) tissue blocks for initial diagnosis were available were retrospectively identified. We studied CD68 and FoxP3 expression by immunohistochemical staining on FFPE biopsy specimen and evaluated correlations of their expression with obtained clinical data, treatment response depending on the upfront ASCT, and survival of the patients. The cut-off value for the expression of CD68+ TAMs and FoxP3+ Tregs were evaluated by the area under curve (AUC) of the receiver operating characteristic (ROC) curve for analysis purposes, and we established cut-offs of 55 cells/high power field (HPF) for CD68 and 15 cells/HPF for FoxP3. We stratified patients based on CD68 and FoxP3 expression according to the cut-off values we determined from the AUC. The overall survival (OS) and progression-free survival (PFS) were plotted using the Kaplan-Meier method and compared using the log-rank test. The Cox proportional-hazards regression model was used in both univariate and multivariate analyses. Results The median age of the patients was 57 (range 33-79) years. The median follow-up duration for survivors was 23.2 months (range, 2.4-128.5). Sixteen (21.1%) patients underwent upfront ASCT, after median 4 (range 2-4) cycles of HD-MTX based chemotherapy. The 2-year OS and PFS rates for all patients were 75.2% and 43.3%, respectively. The patients did not reach median OS, and the median PFS was 17.9 months (95% confidence interval [CI], 9.4-26.4). The median level of expression for CD68+ TAM/HPF was 25 (range, 5-80) and the median level of expression for FoxP3+ Tregs/HPF was 0 (range, 0-68). The difference in OS and PFS between the high and low CD68 groups was significant in the univariate (hazard ratio [HR] = 2.79, 95% CI: 0.97-8.03, P = 0.058 for OS, and HR = 2.17, 95% CI: 1.03-4.58, P = 0.043 for PFS), as well as in the multivariate analysis (HR = 3.71, 95% CI: 1.25-11.02, P = 0.018 for OS, and HR = 4.83, 95% CI: 1.91-12.27, P = 0.001 for PFS). The patients with high CD68 expression exhibited 2-year OS and PFS rates of 42.9%, and 10.0%, respectively, in comparison to 81.5%, and 50.7% for those with low CD68 expression (P = 0.048 for OS, and P = 0.035 for PFS) (Figure 1A, 1B). In a subgroup analysis of 60 patients who did not receive upfront ASCT, high CD68 expression was associated with inferior OS and PFS compared to low CD68 expression (P = 0.014 for OS, and P = 0.016 for PFS) (Figure 1C, 1D). The difference in OS and PFS between the high and low CD68 expression groups in the non-upfront ASCT subgroup (n = 60) was significant in the univariate (HR = 3.63, 95% CI: 1.21-10.88, P = 0.021 for OS, and HR = 2.60, 95% CI: 1.15-5.86, P = 0.021 for PFS) as well as in the multivariate analysis (HR = 4.05, 95% CI: 1.35-12.16, P = 0.013 for OS, and HR = 5.80, 95% CI: 2.25-14.95, P < 0.001 for PFS). However, the OS and PFS in the upfront ASCT cohort (n = 16) were similar between the high and low CD68 expression groups (P = 0.426 and P = 0.848, respectively) (Figure 1E, 1F). There were no differences in OS and PFS according to the expression level of FoxP3 in all patients as well as in subgroup of patients who did not receive upfront ASCT. Conclusion High level of CD68 expression in patients with PCNSL was significantly associated with inferior OS and PFS, especially in non-upfront ASCT treated subgroup of patients. FoxP3 expression level was also not associated with survival in this study. We suggest CD68 as a potential biomarker at initial PCNSL diagnosis and upfront ASCT consolidative strategy might improve survival in PCNSL patients by overcoming negative impact of high CD68 expression. Further validation studies are warranted. Figure 1. Figure 1. Disclosures No relevant conflicts of interest to declare.

Abstract 4149 Background Follicular lymphoma (FL) is the second most common type of lymphoma in Europe and North America. Traditionally, FL has been graded according to the number of centroblasts in the follicles per high power field, to grades 1–3. This grading system, however, does not capture the clinical variability observed in low grade (1-2) FL. The recent WHO classification made it clear that clinically, FL grade 1 and 2 cannot be distinguished, with generally long median survival and recurrent relapses, while grade 3 FL generally has a more aggressive course. Though the clinical prognostic indices such as the FLIPI and the IPI can provide information regarding outcome, they fail to define the biology of the disease. Proliferation index (PI) as determined by Ki-67 immunohistochemistry (MIB-1) has been shown by some to have prognostic value in FL, while others did not find any significance. The PI value in FL is usually determined either by counting cells or by visual estimation. We previously validated the use of quantitative image analysis (QIA) in mantle cell lymphoma, and showed that it is a robust tool for determination of the PI. It is objective and highly reproducible. In this study we aim to evaluate the prognostic significance of PI in a retrospective cohort of pts with FL, grade 1–2. Methods Specimens from 73 FL pts (33 man, 40 women) diagnosed between the years 2001–2008 were identified. Clinical data was obtained by thorough chart review. All specimens were reviewed by an experienced hematopathologist and the grade confirmed. There were 59 grade 1–2 and 14 grade 3 cases. Slides were scanned with the ScanScopeXT (Aperio, Vista, CA) using brightfield imaging at 200x magnification. Follicles were annotated using ImageScope (Aperio) and analyzed using Aperio image analysis software. Inter-follicular areas were not analyzed. Survival analysis was done by the Kaplan-Meier method. Differences between groups were analyzed using chi-square test and Fisher's exact test. Results Median age at diagnosis was 61.8 years. Median follow-up time for all 73 pts was 38.8 m (range 0.95 to 102.3). The median PI for the entire group was 33%. When analyzing grade 1 and 2 separately (no=59), the median was almost the same (32%) (Fig 1). We found no significant correlation between grade and PI in this cohort. Outcome for pts with grade 1–2 disease was analyzed by dividing the pts based on PI: low PI (PI=30%). These two groups did not differ in clinical characteristics including: age, clinical stage, FLIPI score, or the indication for first treatment. Overall survival (OS) (median not reached) and event-free survival (EFS) (median 54.2 m) were not statistically different between the groups. However, time to first therapy (TTFT), was significantly shorter for the high PI group (p=0.01) (Fig 2). The median time to first therapy for FL, grade 1–2, was 3.4 m. Within this group, the time to first treatment was 2.1 m v 31.6 m, for the high and low PI groups, respectively. As the curves indicate, the difference between the groups starts approximately 3–6 months after diagnosis. A similar analysis of FLIPI risk group (low and intermediate v high) failed to demonstrated differences in OS, EFS and TTFT. Conclusions We conclude that PI varies significantly among low grade FL pts. This variability may relate to the clinical course of these pts. We have shown significantly different time to first therapy between the low and high PI groups. We plan to further explore differences in OS and EFS in a larger series of FL pts and confirm the predictive value of PI for TTFT. PI distribution as determined by QIA among all cases of FL. Time to first therapy for low grade FL pts separated to low and high PI. Disclosures: No relevant conflicts of interest to declare.

BACKGROUND: THAL has been used as immunomodulatory therapy alone (Tefferi et. al. Blood 200;96:4007), or with steroids (Mesa et al. Blood2003;101:2534), in patients with primary myelofibrosis (PMF). Clinical benefit has been limited to improvements in cytopenias and/or splenomegaly. However, major histologic remissions of intramedullary manifestations of disease have not typically been observed. The mechanism of action of THAL in MF remains uncertain, and may involve cytokine inhibition or immunomodulation. We have previously observed increased numbers of CD8+ T lymphocytes on marrow trephines of patients who responded to THAL, and hypothesized that alterations in the T-cell distribution and population may provide insight into an immunomodulatory response and correspond to subsequent response in PMF patients. Methods: We analyzed a cohort of PMF patients who had received THAL therapy (alone or in combination with corticosteroids) from our institution and analyzed their clinical course and response to THAL therapy. Marrow trephines were evaluated in a systematic fashion (CYL in a blinded fashion to clinical course) for cellularity and degree of reticulin fibrosis (0–4+). Subsequently, megakaryocytes were assessed for quantitative (increased, decreased, or normal) and qualitative (size (large or small) and presence or absence of morphologic dysplasia). T cells were assessed by immunohistochemical staining for CD8 with an emphasis on number of T cell clusters per trephine, and the pattern of those clusters (large vs. small). Finally, an estimation of numbers of T cells per high power histology field was conducted in 3 distinct areas, and an average per marrow trephine obtained. All histologic features were then compared with PMF prognostic scores, and labs at presentation. Amongst those who received THAL, baseline histology was compared to response and in those patients with serial marrows longitudinal changes were assessed. Results: A cohort of 140 patients with PMF was analyzed, 65 having received THAL along the course of their illness (53 (82%) along with prednisone) and 75 from a comparison group who never received THAL. Patients from both groups had similar PMF prognostic scores and presentations. Comparison of baseline histologic features demonstrated that individuals with a baseline increases in T cell numbers (>100 per high power field) were significantly more likely to subsequently respond to THAL therapy (p Table 1. Comparison of Histologic Features in 140 patients with Myelofibrosis Group Cell Fibrosis ≥3 Meg Numbers Meg Size T Cell clusters T-Cells (per HPF) >100 Tcells (per HPF) Control (N=75) 72% (10–100) 33% 91% Increased 21% Small 4 (0–16) median 80 (13–258) 37% THAL ALL (N=65) 60% (5–95) 62% 79% Increased 38% Small 3 (0–14) median 93 (20–500) 43% -THAL Responders (N=28) 60% (5–95) 56% 82% Increased 22% Small (p=0.05) 5 (0–10) median 113 (45–198) 61% (p A subgroup of patients who received THAL (N=12) had a second post therapy marrow available for analysis for comparison of THAL effects upon the marrow, although the numbers were small, there was no statistically significant differences upon histologic parameters analyzed including cellularity, degree of fibrosis, megakaryocyte morphology, or T cell clustering or numbers. Inclusion of prednisone in the regimen had no impact on histologic prognostic observations. Conclusions: THAL therapy for MF patients can be associated with significant clinical benefit for cytopenias and splenomegaly, and interestingly patients with increased marrow T-cell populations at baseline have a clear increased likelihood of response. Although this analysis is unable to gauge mechanism, it would suggests a greater potential for immunomodulation in patients with a robust T-cell population. Additionally, analysis of megakaryocytes suggests the presence of small megakaryocytes may in contrast represent a more resistant phenotype of disease with more thrombocytopenia and THAL resistance.

Background: The tumor microenvironment is important in the biology of follicular lymphoma (FL). In addition to macrophages, which have been shown to be associated with outcome in FL, mast cells (MCs) have the ability to influence the microenvironment through secretion of numerous biologically active molecules such as growth factors and cytokines/chemokines. Increased numbers of tumor-infiltrating MCs have been suggested as an unfavorable prognostic marker that also supercedes the predictive value of lymphoma-associated macrophages (LAM) in FL patients treated with immunochemotherapy. This study examines the effects of tumor-infiltrating MCs in FL patients at Cleveland Clinic. Design: Tissue microarrays were constructed from 94 newly diagnosed cases of FL from 09/01/1985 to 12/13/2002 that had been previously characterized for CD68+ LAMs and demonstrated that extrafollicular (EF) LAMs (>16.8/high power field) were a poor prognostic marker (Kelley T et al 2007). Immunohistochemistry for tryptase was performed on these microarrays to enumerate MCs per high powered (400x) field. In each core, all available lymphomatous tissue was evaluated (mean of 8.13 hpf/case, range 1–9). The cell counts were compared to the available clinical follow-up data, using overall survival as the measurable outcome. Results: The FL patients (44 women, 50 men) had a mean age at diagnosis of 59 years and a mean follow-up of 95.8 months. The patients were treated heterogeneously; the great majority did not receive immunochemotherapy as initial treatment. MC data was available in 91 patients. The mean MC count was 4.14 MCs/hpf (range 0–17.33). As shown previously, > 16.8 EF LAMs/hpf was a predictor of poor overall survival; hazard ratio of 2.28, 95% CI: 1.22–4.24, P = .009. In the entire cohort, MCs were not associated with overall survival. However, in the subgroup with low LAM (CD68 3.3/hpf) was associated with better overall survival, HR = 0.40, 95% CI: 0.16–0.99, P = .049. Conclusion: In our data set, increased LAMs in FL are associated with poor overall survival. Overall, MCs do not appear to predict outcome; however, in the subgroup of patients with low LAMs, increased MCs are associated with a favorable overall survival. This suggests a complex interaction between cellular constituents of the tumor microenvironment that affects patient outcome. The application of immunochemotherapy or other immunomodulatory therapies may alter these interactions and outcomes. Further work is needed to better understand these interactions and to validate biomarkers in therapy specific contexts.

Introduction: FL is an indolent disease with heterogeneous survival. Transformation is asscoiated with short survival. In solid tumors, tumor angiogenesis is required for tumor growth/metastasis and microvessel density (MVD) has, in some tumor types, shown correlation with poor OS. In FL, the impact of MVD on prognosis is controversial. Recent studies have highlighted the effect of the microenvironment on FL prognosis including ours which showed that high numbers of lymphoma-associated-macrophages (LAM) are associated with adverse outcome. Because LAM have been shown to promote angiogenesis and neoplastic growth, we used a quantitative tumor mapping approach to determine whether the degree of vascularization in FL correlated with adverse outcome. Method: Between 1987 and 1993, 126 pts were enrolled on a phase II study of BP-VACOP with involved region radiotherapy. All pts were treatment naïve, < 61y and had advanced-stage FL. Paraffin blocks of the diagnostic biopsies were available for 105 pts. Images of whole sections stained for CD34 were captured using a cooled CCD camera, a motorized stage, and customized NIH-Image software. The motorized stage allowed for tiling of adjacent microscope fields of view, thereby allowing reassembly of the entire tumor section at high-resolution. The distance from each point in the tissue to the nearest CD34+ pixel was measured and the tumor to vessel distance that encompassed 90% of the tumor (TVD90) calculated. A TMA was immunostained with S100A8/Calgranulin A-specific antibody, a M2-like macrophage marker, to determine a subset of LAM. Univariate & multivariate analyses were used to determine OS and RT. Results: There were 84 evaluable cases. The median follow-up of the living patients was 14.3 y and the estimated 10-y OS and RT were 65% and 20%, respectively. The IPI was predictive of OS (RR =3.4, 95% CI =1.8–6.3, p < 0.001) and RT (RR =3.3, 95% CI =1.3–7.9, p = 0.01). Histologic grade included 66 grade 1, 14 grade 2 and 4 grade 3a. Cases revealed TVD90 of 50.3 to 144 microns (median 78.4) and were divided into quartiles. A subset of 14 cases showed high density of S100A8+ cells (> 5 per high power field). In univariate analysis the quartile (21 cases) with low TVD90 (i.e. high MVD) showed inferior OS (p < 0.001) and high RT (p = 0.012). Cases with high density of S100A8+ LAM also had inferior OS (p = 0.04), but the high density did not predict RT (p = 0.4). A Cox multivariate model with IPI, S100A8+ score and TVD90 showed only IPI and TVD90 to be independent predictors of OS (RR = 2.9, 95% CI = 1.6–5.6, P = 0.001; RR = 2.4, 95% CI = 1.3–4.7, p = 0.009, respectively), but only TVD90 predicted RT (RR = 3.1, 95% CI = 1.2–7.6, p = 0.017). Interestingly, low TVD90 (i.e. high MVD) correlated with increased content of S100A8+ cells (χ2, p < 0.001). Conclusion: These results show that in FL, increased vascularization is independently associated with an adverse clinical outcome and, most importantly, predicts for increased RT. These findings may provide a rationale for trials of antiangiogenic therapy in “aggressive” FLs. Tumor to Vessel Distance (TVD) and Follicular Lymphoma Overall Survival & Risk of Transformation Tumor to Vessel Distance (TVD) and Follicular Lymphoma Overall Survival & Risk of Transformation

High amounts of intra-tumoral reactive macrophages have been reported to be associated with a poor prognosis in patients with follicular lymphoma (FL) (Blood2005;106:2169). However, this study has been retrospectively performed in a patients’ population selected before the monoclonal antibody therapeutic area, and has not been confirmed on independent series. To establish whether the intra-tumoral macrophages count (MC) is definitely able to predict the outcome of FL patients, we analysed both immunohistochemical CD68 expression and clinical outcome in patients included in the GELA-GOELAMS FL-2000 trial (ASH-2004; ASCO-2006). Patients between 18 and 75 years of age with high tumor burden FL were randomized to receive either CHVP-I or R-CHVP-I. Among the 358 patients included on the basis of histologically proven FL after reviewing by 3 hematopathologists, 194 had available biopsy specimen. Clinical characteristics of those 194 patients were not different of those of the whole population and median time to event was 43.1 months in 92 pts receiving CHVP-I arm and 54.5 months in the 102 pts receiving R-CHVP-I. Slides were stained with the anti-CD68 KP1 antibody using a standard IHC procedure, then scored on high power field (hpf) (× 400 magnification) by 3 different pathologists. Analysis of the relative risk of event according to the intra-follicular MC (IF-MC) led us to determine a single cut-off point to categorize the covariate, the best cut-off point being estimated to be 10 macrophages/hpf. The IF-MC was equal or less than 10 intra-follicular KP1+ macrophages/hpf in 53 patients, and more than 10 in 141 patients. With this cut-off of 10, a low MC was significantly associated with a better EFS for all patients (p=0.011). However, this effect was predominantly observed in patients that received CHVP-I (p=0,012, OR) but not in those receiving R-CHVP-I (p=0.15).Multivariate analysis adjusting for 10 IF-MC cut-off (p

Background: Regulatory T-cells (Tregs), a CD4+, CD25+, FOXP3+ subset, appear to modulate the tumor microenvironment in lymphoma. Using FOXP3 expression as a marker of Tregs, single series have suggested that higher numbers of tumor infiltrating FOXP3+ cells are associated with improved survival in both follicular lymphoma and classical Hodgkin lymphoma (cHL). However, neither of these findings has been confirmed by separate studies and, in fact, contradictory data exists. We sought to evaluate potential correlations between the number of FOXP3+ cells in the tumor microenvironment and patient outcome. Methods: 98 previously untreated cHL patients, biopsied and treated with curative intent at the Cleveland Clinic, were studied. FOXP3 was assessed by immunohistochemistry (Abcam) and quantitated as the number of positive cells per high power field (hpf), averaged over 5 fields. Histology was evaluated and medical records were reviewed for clinical features (including gender, age, stage, B symptoms, bulk, number of sites and IPS), lymphoma progression and death. Results: The mean number of FOXP3+ cells (±standard deviation) was 41.7 ± 33.9 cells/hpf, and the median number was 36.4 cells/hpf (range 0–152.0 cells/hpf). Higher numbers of FOXP3+ T-cells were associated with age25.0 FOXP3+ cells/hpf compared to 81% ± 6% for patients with Conclusion: FOXP3+ T cells are a favorable prognostic marker in cHL and reflect the importance of the immune response thus providing rationale for the use of immunomodulatory therapies in these patients. Failure Free Survival Failure Free Survival Overall Survival Overall Survival

Mantle cell lymphoma (MCL) is a malignant lymphoma associated with a relatively aggressive clinical course and a median overall survival time of 3–4 years. We investigated immunohistochemically the expression of the apoptotic marker caspase 3 in relation to the clinical course. Biopsy specimen from 84 untreated patients enrolled in two multicenter prospective trials were investigated immunohistochemically with monoclonal antibodies against CD20, CD5, CD3, CD23, cyclin D1, Caspase3. The Caspase3 expression was analyzed in three groups: less than 1 positive cell per high power field (HPF), more than 1 positive cell per HPF and more than 2 positive cells per HPF. The expression was compared with the overall survival data analysed according to Kaplan and Meier. In 75 cases the caspase 3 staining could be analyzed. The caspase 3 expression had a range of 0.1–4.7 positive cells per HPF (median value:1.1, mean:1.3). Patients with mantle cell lymphoma that had less than 1 caspase 3 positive cell per HPF (33 cases) had a median overall survival time of 48 months compared to 27 months for patients with more than 1 positive cell per HPF (24 cases) and 15 months for more than 2 positive cells per HPF (18). The Kaplan-Meier analysis showed a significant difference in the overall survival time between these groups (p

Gene expression profiles revealed that proliferation associated genes are important prognostic factors in the clinical outcome in mantle cell lymphoma (MCL). Beside this well accepted markers also analysis of apoptotic proteins are now under investigation. We investigated immunohistochemically the expression of the apoptotic marker NF-kB in relation to the clinical course in 89 patients enrolled in two multicenter prospective trials. Biopsies were recut and stained with mandatory antibodies (CD20, CD5, CD3, CD23, cyclin D1) and NF-kB. The NF-kB expression was analyzed in three groups: negative, cytoplasmatic positive and nuclear positive (more than 1/HPF). The expression was compared with the overall survival data analyzed according to Kaplan and Meier. In 13 cases a negative NF-kB staining was detected. Fifty-seven cases were positive only in the cytoplasma and in 17 cases more than 1 cell per high power field showed nuclear activity. Patients with mantle cell lymphoma that had negative and cytoplasmatic positive expression had a median overall survival time of 35.7 months compared to 22.4 months for patients with a nuclear NF-kB expression. The Kaplan-Meier analysis showed a significant difference in the overall survival time (p=0.0121). The immunohistochemical detection of NF-kB in mantle cell lymphoma is possible and a tool to identify patients with a poor prognosis.

Background: Abnormally increased tumor associated neovasculature plays an important role in tumor progression in solid tumors and hematological malignancies. In multiple myeloma (MM) bone marrow angiogenesis, measured in terms of microvessel density (MVD), has prognostic value, and appear to increase with disease progression from monoclonal gammopathy of undetermined significance to relapsed MM. We have shown that MVD has prognostic value in patients with newly diagnosed MM including patients undergoing upfront high dose therapy and stem cell transplantation, but comparison with other known prognostic factors has been limited by sample size. It is also not known whether the MVD adds prognostic value once the recently described International Staging System (ISS) is applied. The goal of this study was to determine the prognostic effect of bone marrow MVD in newly diagnosed MM relative to the ISS and other known prognostic factors. Methods: We studied 400 patients with newly diagnosed MM seen at the Mayo Clinic between March of 1988 and December 2001. Sections from bone marrow biopsy blocks from the time of initial diagnosis were studied by immunohistochemistry using antibodies against CD34 antigen to high light the endothelial cells. Under low power (100X); three areas with maximum number of microvessels (hotspots) were identified. Each of these areas was further studied at 400X magnification and the number of microvessels per high power field counted. The average of the three readings was taken as the MVD for the sample. In addition, the samples were graded as low, intermediate or high by using a visual estimate as previously described. Additional clinical data was extracted from medical records. Some of the patients have been included in previous studies related to angiogenesis. Results: A total of 400 patients in whom a bone marrow biopsy from within 30 days of diagnosis was available were studied. The pts were followed for a median of 37 months (Range: 1 month to 16.5 years) and 318 pts (80%) had died at the time of this analysis. The median MVD for the entire group was 14.7 (Range 0–168). The median overall survival for the three groups according to the MVD grade was lower for the high grade group (31.9 months) compared to the intermediate grade group (37.2 months) and the low grade group (not reached); P = 150 X 106/L), and plasma cell labeling index (=1). In a multivariate analysis using these variables and MVD as a continuous variable, high MVD and the ISS staging system were significantly associated with poorer survival (Table). Conclusion: In this large group of pts with newly diagnosed MM, we confirm the prognostic value of increased bone marrow angiogenesis. We examined the MVD as a continuous variable in the multivariate analysis for a closer evaluation of this measure in this comparison. More importantly, the prognostic value appears to be independent of the ISS and other major prognostic factors. The resultsof this study reinforces the biological relevance of this finding in MM. HR 95% CI P value MVD 1.006 (1.001, 1.011) 0.0279 ISS Stage I 0.37 (0.25, 0.56)

We studied expression of CD34 positive cells in the bone marrow trephines of two groups of patients with CML treated with imatinib. In the first group (n=17), patients went on to achieve sustained complete cytogenetic response during their disease course. In the second group (n =33) complete cytogenetic response was not achieved and patients later developed either accelerated or blastic phase disease. We studied paraffin embedded bone marrow trephines stained with the monoclonal antibody QBEND10 during an eighteen month period following initiation of imatinib therapy and recorded the highest CD34 positive cell counts and aspirate blast counts in three six-month intervals. CD34 positive cells were evaluated using three counting methods. The first method, established in our department for several years, expressed the percentage of CD34 positive cells vs. all nucleated cells in a minimum 500-cell count (% CD34+/500 cells). The second recorded the total number of CD34 positive cells in ten high power fields (CD34+/10hpf) using x 40 magnification /numerical aperture 0.65 and the third method utilised the highest number of CD34 positive cells expressed in a single high power field (CD34+/hpf), the value being taken from the previous CD34+/10 hpf count. We hypothesised that the overall CD34 positive cell burden and absolute CD34 positive numbers would be more important in predicting the likelihood of complete cytogenetic response and this would be better assessed by volumetric evaluation than percentage calculation relative to other nucleated cells. CD34 expression in endothelial cells was taken as a positive internal control. We aimed to establish which variable: % blast count, % CD34+/500 cells, CD34+/10hpf and CD34+/hpf provided better correlation with subsequent cytogenetic response to treatment. Values for each variable were dichotomised at the following levels; aspirate blast count 5, % CD34+/500 cells < 2 and > 2%, CD34+/10hpf < 60 and > 60 cells and CD34+/hpf < 10 and > 10 cells. Univariate analysis was performed on samples within the first six months, seven to twelve months and thirteen to eighteen months. Only CD34+/10hpf showed significant discriminatory ability throughout the entire eighteen month period (Pearson’s Chi-square 2-sided sig: 0.031, 0.003 and 0.007 in the first six months, seven to twelve months and thirteen to eighteen months samples respectively). By logistic regression, we confirmed that CD34+/10hpf had the best correlation with cytogenetic response during the first year following initiation of imatinib therapy. Larger studies are required to confirm these initial findings and to establish volumetric assessment of CD34 positive cells as a predictor for cytogenetic response.