A morphometric study of bone marrow angiogenesis in hairy cell leukaemia with clinicopathological correlations

A morphometric study of bone marrow angiogenesis in hairy cell leukaemia with clinicopathological correlations
of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  A morphometric study of bone marrow angiogenesis in hairycell leukaemia with clinicopathological correlations Penelope Korkolopoulou, 1 Despina A. Gribabis, Nikolaos Kavantzas, 1 Maria K. Angelopoulou, 2 Marina P. Siakantaris, 2 Efstratios Patsouris, 1 Athina Androulaki, 1 Irene Thymara, 1 Styliani I. Kokoris, 2 Maria C. Kyrtsonis, 2 Christos Kittas 2 and Gerassimos A. Pangalis 21 Department of Pathology and   2 Haematology Section, First Department of Internal Medicine, School of Medicine, LaikonGeneral Hospital, National and Kapodistrian University of Athens, Athens, GreeceReceived 27 November 2002; accepted for publication 27 May 2003 Summary.  Bone marrow angiogenesis has recently beenimplicated in the pathophysiology and course of varioushaematological malignancies. Little is known, however,about the significance of this phenomenon in hairy cellleukaemia (HCL). We evaluated various morphometriccharacteristics of microvessels, highlighted by means of anti-CD34 immunohistochemistry, in the bone marrow of 44patients with typical HCL, before and after treatment withinterferon- a  (IFN- a ). Overall, bone marrow from 103 HCLpatients and 20 controls was examined. Microvessel density(MVD) and several size- and shape-related parameters werequantified in the region of most intense vascularizationusing image analysis. MVD, size-related parameters and thepercentage of branching microvessels were higher in HCLthan in controls. Likewise, perimeter counts were higher inpartial/non-responders than in complete responders.Achievement of complete response was accompanied bysmaller calibre microvessels. IFN- a  induced a decrease inMVD and branching values in cases with diffuse marrowinvolvement. In univariate analysis, progression-free survi-val was adversely affected by MVD, branching and majoraxis length. Multivariate analysis indicated that MVD/branching independently affected progression-free survivaland the likelihood of complete response. Our data suggestthat the generation of bone marrow microvessels indicatedan increased risk of progression and IFN- a  treatment failurein HCL. Furthermore, the prognostic significance of angio-genesis requires the concomitant assessment of MVD and thecomplexity of the microvascular network. Keywords:  hairy cell leukaemia, CD34, angiogenesis,interferon- a .The viability of solid tumours is sustained by the sprouting of new vessels from pre-existing vasculature, termed angiogen-esis (Folkman, 1995). The advent of the tumour into avascular phase entails a shift in the balance towards theangiogenic stimulators, endowing tumour cells with theability to recruit blood vessels (Hanahan & Folkman, 1996;Dor & Keshet, 2000). A plethora of investigations haveverified that the conversion to an angiogenic phenotypeheralds the onset of biological aggressiveness in a variety of malignancies (Srivastava  et al , 1988; Brawer  et al , 1994;Yamazaki  et al , 1994; Hanahan & Folkman, 1996; Pavlopo-ulos  et al , 1998; Zetter, 1998; Korkolopoulou  et al , 2001a).Research into the angiogenic aspects of haematologicaloncology has recently been stimulated by perceived parallelswith solid tumour angiogenesis. Indisputable evidence hasaccumulated (Mangi & Newland, 2000; Keyhani  et al ,2001; Ribatti  et al , 2001) indirectly implicating bonemarrow angiogenesis in the pathophysiology and courseof various haematological malignancies (Perez-Atayde  et al ,1997; Aguayo  et al , 1999, 2000a,b; Mangi & Newland,2000; Padro¨  et al , 2000; Salven  et al , 2000;De Bont  et al , 2001; Korkolopoulou  et al , 2001b; Ribatti et al , 2001; Korkolopoulou  et al , 2003).Published information regarding angiogenesis in chroniclymphoid leukaemias has been centred upon B-cell chroniclymphocytic leukaemia (B-CLL), in which bone marrowmicrovessel density (MVD) and the production of potentangiogenic molecules, notably vascular endothelial growthfactor (VEGF) and basic fibroblast growth factor (bFGF),correlate withclinicalstageandoutcome (Baban et al ,1996;Menzel  et al , 1996; Konig  et al , 1997; Molica  et al ,1999; Aguayo  et al , 2000b; Chen  et al , 2000; Kini  et al ,2000; Kay  et al , 2001; Molica, 2001). Very little is knownabout the angiogenic basis of hairy cell leukaemia (HCL), an Correspondence: Dr P. Korkolopoulou, 73 Vassileos Pavlou Street,GR-15452 Psychico, Athens, Greece. E-mail: British Journal of Haematology,  2003,  122 , 900–910 900    2003 Blackwell Publishing Ltd  allied B-cell chronic lymphoid leukaemia, although tworecent reports support the presence of increased marrowvascularity in a small number of HCL patients (Kini  et al ,1999; Pruneri  et al , 2003).In the present study, we evaluated multiple morphomet-ric characteristics of microvessels, in addition to MVD, inHCL patients in relation to clinicopathological parametersand response to therapy. Such information might provevaluable in clarifying certain aspects in the biology of thedisease and would provide insight into the potential clinicalsignificance of angiogenesis in these patients.PATIENTS AND METHODS Patients.  Patients ( n  ¼  44) with typical HCL diagnosedbetween 1982 and 2000 were enrolled in this study.Diagnosis was based on standard clinical, morphological,immunological, histological and immunohistological fea-tures (reviewed by Pettitt  et al , 1999; Tallman  et al , 1999)and was independently confirmed by three haematologists.There were 33 men and 11 women with a median age of 53 years (range 31–76 years). The clinical, laboratoryand bone marrow histological data of the patients aresummarized in Table I. All patients had been treated with a -interferon (Intron-A; Schering, Kenilworth, NJ, USA) witha daily dose of 3 MU subcutaneously (s.c.) for 12 (range6–18) months. Splenectomy was performed in 15 cases.Patients were evaluated clinically, with laboratory tests andbone marrow examination at 3, 6 and 12 months aftertreatment initiation and then every 6 months.Patients were considered evaluable for response at12 months. The response criteria used are based on theprinciples of the Second International Hairy Cell LeukaemiaConference (Catovsky  et al , 1987) and those stated byRatain  et al  (1988). Complete response was defined asnormalization of peripheral blood counts (haemoglobin ‡ 12 g/dl, platelets  ‡ 100  ·  10 9 /l and absolute neutrophilcount  ‡ 1 Æ 5  ·  10 9 /l), disappearance of palpable disease andabsence of discernible hairy cells in the bone marrow byconventional histological examination; however, thepresence of <5% hairy cells by immunohistologicalexamination (L26/DBA-44) was allowed (Ratain  et al ,1988). Partial response was defined as normalization of peripheral blood counts, disappearance of palpable diseaseand >5% residual hairy cells in the bone marrow. Patientsnot meeting the criteria for complete or partial responsewere categorized as non-responders. Twenty-two patients(50%) were judged to have completely responded, 11 (25%)had partially responded and 11 (25%) had not responded.Patients were followed up for a median period of 124 months (range 12–240 months). Within this period,only one patient died of disease-related causes, whereas 15patients progressed after a median period of 57 months(range 14–157 months). Of the remaining patients, 29 hadhistologically stable disease for a median period of 28 months (range 14–186 months). Features of progressivedisease included worsening of at least one parameter of theperipheral blood (i.e. anaemia, neutropenia or thrombocy-topenia) along with an increasing proportion of hairy cellsin immunohistological examination of the bone marrow(Filleul  et al , 1994). Selection of CD34 as the endothelial marker.  During theinitial phase of the present investigation, we compared thestaining performances of antibodies to the endothelialantigens CD34 and factor VIII (FVIII) in 15 randomlyselected cases. The values for the various angiogenicparameters obtained with both antibodies strongly correla-ted with each other, the correlation coefficient ranging from0 Æ 981 to 0 Æ 996. However, staining for FVIII was inferior tothat obtained with CD34, because of increased backgroundstaining, strong staining of megakaryocytes and fainterlabelling of endothelial cells. Therefore, the results refer onlyto CD34 staining (Kini  et al , 1999, 2000). Processing of bone marrow specimens and immunohistochem-ical staining for CD34.  Bone marrow specimens were fixed inbuffered formalin, decalcified with EDTA/HCl and embeddedin paraffin. All biopsies were reviewed to assess uniformlythe degree of immunohistologically (i.e. by means of L26/DBA44 antibodies) determined hairy cell infiltration, thepattern of bone marrow infiltration (Pangalis  et al , 1987; Table I.  Clinicopathological and laboratory characteristics of 44 HCL patients at diagnosis.Characteristic  n  Median RangeAge (years) 44 53 31–76Gender (male/female) 33/11Palpable spleen size (cm) below costal margin (yes/no) 27/17 4 0–25Blood counts 43Haemoglobin (g/dl) 10 Æ 8 6 Æ 5–14 Æ 3Platelets ( · 10 9 /l) 63 Æ 0 15 Æ 0–166 Æ 0WBC count ( · 10 9 /l) 3 Æ 3 1 Æ 2–20 Æ 6Circulating hairy cells ( · 10 9 /l) 0 Æ 8 0 Æ 0–16 Æ 5Absolute neutrophil count ( · 10 9 /l) 0 Æ 8 0 Æ 1–2 Æ 5Bone marrow histologyCellularity (%) 70 30–98Degree of infiltration by hairy cells (%) 72 Æ 5 10–99Pattern of involvement (diffuse/interstitial/focal) 17/26/1 Angiogenesis in Hairy Cell Leukaemia  901   2003 Blackwell Publishing Ltd,  British Journal of Haematology  122 : 900–910  Katayama, 1988) and the percentage of residual fat cells.Serial sections (4  l m thick) of each sample were alsoprocessed for immunohistochemical identification of micro-vascular endothelial cells with a mouse monoclonal anti-body against CD34 (Clone HPCA-1; Becton Dickinson, San Jose, CA, USA). The antibody was applied at a dilution of 1:80 for 1 h. Before staining, slides were incubated fourtimes for 5 min in citrate buffer, pH 6 Æ 0, at 750 W in amicrowave oven (Shi  et al , 1991). Application of theprimary antibody was followed by the standard three-stepstreptavidin peroxidase technique (Hsu  et al , 1981). Allspecimens were treated using identical procedures. Negativecontrols (i.e. sections in which the primary antibody wassubstituted with non-immune mouse serum) were alsostained in each run.Overall, 103 bone marrow biopsies taken from 44 HCLpatients were evaluable for CD34 immunohistochemistry.These included 44 pretreatment biopsies and 59 post-treatment biopsies at 6 months (two cases), 12 months (19cases) and up to 170 months (38 cases) from diagnosis. In12 patients, more than two biopsies (three to five) wereavailable for the study of microvessels. Control marrowsfrom 20 subjects with no evidence of marrow disease,performed for diagnostic purposes as part of a stagingprocedure for Hodgkin’s disease (13 cases), non-Hodgkin’slymphomas (five cases) and solid tumours (two cases) werealso processed for immunohistochemical staining withCD34. Controls were age and sex matched with HCL cases. Microvessel counting and image analysis.  The method hasbeen described in detail previously (Korkolopoulou  et al ,2001a,b). Briefly, CD34-stained slides were scanned at100  ·  magnification (Padro¨  et al , 2000) and then at 250  · magnification to identify the vascular ‘hot-spot’ (Kini  et al ,2000). Selection of hot-spot has been adopted as a standardprocedure for the study of angiogenesis in both solidneoplasms and haematological oncology (Hlatky  et al ,2002). It is thought that such areas of increased concen-tration of microvessels may represent the emergence of aneoplastic clone with a higher angiogenic potential (Rak et al , 1995). The hot-spot was stored as a 24-bit BMP file.For each countable microvessel, the outline was traced(Fig 1), and various morphometric parameters relating tomicrovessel calibre (major and minor axis length, perimeter,area and Feret diameter) and shape (compactness and shapefactor) were estimated (Fox & Ulrich, 1995). The total countof microvessels per optical field (MVD) and the number of vascular ramifications per 100 vessel sections, as anexpression of the complexity of the microvascular network,were also recorded (Pavlopoulos  et al , 1998). Statistical analysis.  To confirm that the above countingtechnique was representative of bone marrow angiogenesis,we counted the microvessels in three random fields in 10randomly selected HCL cases and found that the meanvalues of angiogenic parameters correlated with those of ahot-spot ( r   ¼  0 Æ 871–0 Æ 993). MVD was not related to thesize of the bone marrow biopsy ( P  > 0 Æ 10), arguing againsta significant bias in hot-spot selection caused by the lowvariation in specimen length. Differences in angiogenesisbetween controls and the various groups of HCL or beforeand after interferon- a  (IFN- a ) were examined by theindependent samples  t -test, one-way analysis of variance( anova ) and paired samples  t -test. The relationship of microvascular parameters to clinicopathological character-istics and the effect of the latter on microvascular changesafter IFN- a  were tested with the independent samples t-test,chi-square test, Fisher’s exact test and Pearson’s correlationcoefficient. Associations among microvascular parameterswere tested with Pearson’s correlation coefficient.Given that, during the follow-up period, only one patientdied of disease-related causes, survival analysis was restric-ted to progression-free survival. The effect of variousparameters on progression was assessed by plottingKaplan–Meier survival curves and comparing groups usingthe log-rank test as well as with multivariate analysis usingthe stepwise Cox’s regression model. Numerical variableswere categorized on the basis of the median value roundedto the nearest 10%. The independent predictive power of each variable in terms of the likelihood of response to IFN- a was evaluated with logistic regression. Given that therewere significant inter-relations among the microvascularparameters (see Results), the set of microvascular variableswas subjected to factor analysis using the principal compo-nent extraction method. Multivariate analysis included theestimated factor scores in continuous form (instead of thesrcinal microvascular parameters) along with the remain-ing variables.RESULTS Comparison of morphometric variables between controlsand HCL MVD and size-related parameters [area, minor and majoraxis length, perimeter and Feret diameter (the measureddistance between theoretical parallel lines that are drawntangent to the particle profile and perpendicular to theocular scale)] were higher in HCL than in controls, althoughFeret diameter attained a borderline significance in thisregard (Table II). Shape factor and compactness were notsignificantly different between the two groups. Controls alsoshowed a lesser degree of vessel branching than HCL( P  ¼  0 Æ 045, Fisher’s exact test). These differences translatedto a large number of large-calibre, often arborizing, vesselsin HCL bone marrow, as opposed to a few simple andstraight microvessels in controls (Fig 1). Within the controlgroup, none of the microvascular parameters correlatedwith cellularity. Association of morphometric variables at diagnosiswith clinicopathological characteristics of HCL patients None of the morphometric variables was related to patients’age, gender, spleen size, haemoglobin level, platelet count,white blood cell (WBC) count, the percentage of circulatinghairy cells or the degree of bone marrow involvement andthe absolute neutrophil count ( P  > 0 Æ 10). Although norelationship existed between microvascular parameters andmarrow cellularity, cases with diffuse involvement displayedhigher microvessel counts ( P  ¼  0 Æ 004), a higher degree of arborization ( P  ¼  0 Æ 013) and marginally elevated values of  902  P. Korkolopoulou et al   2003 Blackwell Publishing Ltd,  British Journal of Haematology  122 : 900–910  Fig 1.  (A) Immunohistochemical staining of bone marrow endothelial cells for CD34 in a case of HCL. (B) The same field when the outline of each vessel has been traced; the grey layer represents the section areas of each vessel. The white perpendicular lines correspond to the majorand minor axes of each microvessel. Two branching vessels can be discerned. The MVD in this case was 22. Original magnification  · 250. Angiogenesis in Hairy Cell Leukaemia  903   2003 Blackwell Publishing Ltd,  British Journal of Haematology  122 : 900–910  minor axis length ( P  ¼  0 Æ 061) than those with interstitialor focal pattern (independent samples  t -test). Correlations among the various morphometric variables in HCL Strong positive correlations existed among vessel calibreparameters (area, major axis length, minor axis length,perimeter and Feret diameter), with the correlation coeffi-cients ranging from 0 Æ 816 to 0 Æ 993 ( P  < 0 Æ 001). Likewise,branching values correlated positively with MVD( r   ¼  0 Æ 763,  P  < 0 Æ 001). As expected, shape factor andcompactness were inversely correlated ( r   ¼  ) 0 Æ 893, P  < 0 Æ 001). Effect of IFN- a  on bone marrow microvascular parameters To study the effect of treatment with IFN- a  on bone marrowangiogenesis, we first compared the microvascularparameters at diagnosis with those after treatment. Overall,no statistically significant difference could be substantiated( P  > 0 Æ 10, paired  t -test). However, when patients werestratified into two groups according to response (completeversus partial or no response), post-treatment bone mar-rows in remission displayed lower perimeter counts thanthose in partial or no remission (independent samples  t -test, P  ¼  0 Æ 038, Fig 2). However, microvessel counts and mostvessel calibre parameters (area, major and minor axislength, perimeter) in post-treatment bone marrows inresponse remained higher than those of control bonemarrows (Table II, Figs 2 and 3). Also, the post-treatmentvessel calibre in non-responders tended to increase (pairedsamples  t -test,  P  ¼  0 Æ 032 for area, perimeter and Feretdiameter,  P  ¼  0 Æ 020 for minor axis length,  P  ¼  0 Æ 050for major axis length) as opposed to responders whose Table II.  Evaluated morphometric parameters in controls, pretreatment HCL and post-treatment HCL bone marrows, stratified by response.Controls( n  ¼  20)  P -value*HCLat diagnosis( n  ¼  44)HCLin completeremission( n  ¼  21)  P -value  HCL inpartial/noremission( n  ¼  22)  P -value  Microvessel count(per optical field)4 Æ 77 ( ±  1 Æ 68) <0 Æ 001 12 Æ 94 ( ±  1 Æ 93) 13 Æ 03 ( ±  1 Æ 85) < 0 Æ 001 11 Æ 26 ( ±  1 Æ 71) <0 Æ 001Major axis length ( l m) 34 Æ 62 ( ±  1 Æ 64) 0 Æ 029 43 Æ 59 ( ±  1 Æ 38) 43 Æ 86 ( ±  1 Æ 40) 0 Æ 049 51 Æ 99 (1 Æ 50) 0 Æ 001Minor axis length( l m) 14 Æ 22 ( ±  1 Æ 47) 0 Æ 013 17 Æ 71 ( ±  1 Æ 33) 18 Æ 70 ( ±  1 Æ 36) 0 Æ 007 21 Æ 11 ( ±  1 Æ 40) <0 Æ 001Perimeter ( l m) 77 Æ 22 ( ±  2 Æ 14) 0 Æ 008 111 Æ 52 ( ±  1 Æ 36) 106 Æ 40 ( ±  1 Æ 42) 0 Æ 025 136 Æ 05 ( ±  1 Æ 49) <0 Æ 001Feret diameter ( l m) 22 Æ 12 ( ±  1 Æ 41) 0 Æ 073 25 Æ 76 ( ±  1 Æ 34) 26 Æ 13 ( ±  1 Æ 31) 0 Æ 086 30 Æ 39 ( ±  1 Æ 39) 0 Æ 001Area ( l m 2 ) 391 Æ 05 ( ±  2 Æ 04) 0 Æ 012 620 Æ 67 ( ±  1 Æ 89) 623 Æ 94 ( ±  1 Æ 79) 0 Æ 028 911 Æ 97 ( ±  2 Æ 17) <0 Æ 001Shape factor 0 Æ 60 ( ±  1 Æ 33) 0 Æ 445 0 Æ 62 ( ±  1 Æ 12) 0 Æ 63 ( ±  1 Æ 13) 0 Æ 175 0 Æ 62 ( ±  1 Æ 11) 0 Æ 397Compactness 21 Æ 58 ( ±  1 Æ 33) 0 Æ 344 22 Æ 73 ( ±  1 Æ 16) 22 Æ 31 ( ±  1 Æ 22) 0 Æ 587 23 Æ 37 ( ±  1 Æ 20) 0 Æ 193Values are presented as geometric means (± standard deviation, SD).*Independent samples  t -test between controls and HCL at diagnosis. Post hoc  comparisons with the least significant difference (LSD) test between post-treatment HCL in   complete remission or in   partial/noremission and controls. Fig 2.  Perimeter counts in controls, pretreatment HCL, post-treat-ment HCL in no/partial remission (NR/PR) and complete remission(CR). Pretreatment HCL displayed elevated perimeter counts com-pared with controls. The same applied to partial/non-responderscompared with complete responders. Fig 3.  Microvessel counts in controls, pretreatment HCL, post-treatment HCL in no/partial remission (NR/PR) and completeremission (CR). Microvessel counts were higher in pre- and post-treatment HCL compared with controls. There was no differencebetween complete and partial/non-responders. 904  P. Korkolopoulou et al   2003 Blackwell Publishing Ltd,  British Journal of Haematology  122 : 900–910
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks