American Journal of Pathology, Vol. 150, No. 3, March 1997
`Copyright C) American Society for Investigative Pathology
`
`Short Communication
`Spectrum of Tumor Angiogenesis in the Bone
`Marrow of Children with Acute Lymphoblastic
`Leukemia
`
`Antonio R. Perez-Atayde, Stephen E. Sallan,
`Usha Tedrow, Susan Connors,
`Elizabeth Allred, and Judah Folkman
`From the Departments of Pathology and Surgery and the
`Division of Hematology and Oncology, Children s Hospital,
`and the Department of Pediatrc Oncology, Dana Farher
`Cancer Institute, Harvard Medical School,
`Boston, Massachusetts
`
`It has been shown that solid tumors progress in
`concert with an induction oftumor angiogenesis.
`It is not known, however, whether a similarphe-
`nomenon occurs in leukemia. Angiogenesis was
`characterized immunohistochemicaly by factor
`VIII staining ofbone marrow biopsies and quan-
`tfifed by assessment ofmicrovessel density using
`previously described techniques. We evaluated
`bone marrow biopsies from 40 children with
`newly diagnosed, untreated acute lymphoblastic
`leukemia. In 22 ofthepatients, we also evaluated
`angiogenesis after the completion of remission
`chemotherapy.
`induction
`Control specimens
`were obtainedfrom children undergoing staging
`evaluations at the time of diagnosis of solid tu-
`and lymphomas.
`mors
`Microvessels
`were
`counted throughout the entire core specimen in
`consecutive X200fields, and a median count per
`field (cpf) was calculated. In addition, the num-
`ber of microvessels in the single X200field with
`the highest microvessel density was designated
`as the "hot spot." Biopsies from children with
`leukemia andfrom controls showed median mi-
`crovessel densities of 42 and 6 counts perfield,
`respectively (P ' 0.0001). Microvessel density of
`the hot spots ofleukemia specimens and controls
`were also significantly different, 51 and 8, re-
`spectively (P < 0.0001). A computer-aided three-
`
`dimensional reconstruction model of bone mar-
`row vascularity showed a complex, arborizing
`branching ofmicrovessels in leukemic specimens
`compared with single, straight microvessels
`without branching in controls. Urinary basic fi-
`broblast growth factor, a potent angiogenicfac-
`tor, was measured in 22 of the children with
`newly diagnosed leukemia and in 39 normal, age-
`matched controls.
`Urinary basic fibroblast
`growth factor levels were increased in aUl 22 pa-
`tients before treatment, were variable during in-
`duction chemotherapy, and demonstrated statis-
`tically insignificant decreases at the time of
`complete remission. These findings suggest that
`leukemia ceUs induce angiogenesis in the bone
`marrow and that leukemia might be angiogene-
`sis dependent and raise the possibilityfor a role
`of antiangiogenic drugs in the treatment of leu-
`kemia. (AmJPathol 1997, 150:815-821)
`
`It is well established that the viability and growth of
`solid tumors is dependent on angiogenesis (neovas-
`cularization).12 Expansion of solid tumors beyond
`the size of a few millimeters in diameter requires the
`induction of new capillaries, a phenomenon medi-
`ated by angiogenic molecules released by tumor
`cells and activated macrophages. It has also been
`shown that the degree of angiogenic activity in a
`given tumor directly correlates with its growth, bleed-
`ing tendency, and potential for metastases; the more
`
`Supported in part by grant CA68484 from the National Institutes of
`Health.
`Accepted for publication November 18, 1996.
`Address reprint requests to Dr. Antonio R. Perez-Atayde, Depart-
`ment of Pathology, Children's Hospital, 300 Longwood Avenue,
`Boston, MA 02115.
`
`815
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 001
`
`

`

`Perez-Atayde et al
`816
`AJPMarch 1997, Vol. 150, No. 3
`
`active the angiogenesis, the greater the probability
`of metastases.1-3
`The angiogenic activity of hematopoietic neo-
`plasms (so-called liquid tumors) has not yet been
`is not known whether leukemic
`demonstrated. It
`cells, which usually grow in the absence of a visible
`connective tissue stroma and circulate in the blood
`and body fluids, are also dependent on neovascu-
`larization to proliferate and expand.
`We therefore studied the extent of angiogenesis in
`acute lymphoblastic leukemia, the most common
`type of childhood leukemia, its modulation with early
`and late chemotherapy, and its correlation with prog-
`nosis. In addition, we investigated the urine levels of
`the angiogenic peptide basic fibroblast growth fac-
`tor (bFGF) in these children and the changes with
`chemotherapy.
`
`Materials and Methods
`To study the response of angiogenesis to treatment,
`we evaluated 61 bone marrow biopsies from 40 chil-
`dren with acute lymphoblastic leukemia. All biopsies
`were obtained for histological diagnosis at presen-
`tation (day 0). In 15 of these children additional serial
`biopsies were available for study; those were usually
`obtained on days 3 (15 biopsies) and 31 (3 biopsies)
`after the initiation of chemotherapy. Seven additional
`children had bone marrow biopsies only on day 31,
`so that comparison of microvessel counts with day 0
`biopsies could not be done. Of these 10 children
`with day 31 biopsies, 6 were in complete and 4 in
`partial remission. Twenty seven of the 40 children
`with acute lymphoblastic leukemia were retrospec-
`tively selected, because they had long follow-up pe-
`riods of more than 10 years, to investigate a possible
`correlation of the degree of angiogenesis with prog-
`nosis. Of these children, 18 had relapsed and were
`either dead with disease or were rescued with bone
`marrow transplantation, and 9 had no evidence of
`relapse and were currently alive and well without
`evidence of leukemia.
`To establish controls, we studied 10 bone marrow
`biopsies of children with various neoplastic disor-
`ders that were obtained as part of their clinical stag-
`ing. Four of the 10 had solid tumors without bone
`marrow involvement (malignant peripheral nerve
`sheath tumor, neuroblastoma, rhabdomyosarcoma,
`and Wilms' tumor), and 6 had lymphoma without
`bone marrow involvement (4 Hodgkin's disease and
`2 lymphoblastic lymphoma).
`One 4-,tm-thick section from each paraffin-em-
`bedded bone marrow biopsy was stained with anti-
`
`factor VIII-related antigen to highlight endothelial
`cells (Dako Polyclonal; Dako, Santa Barbara, CA)
`using a standard immunoperoxidase technique.
`Control bone marrow biopsies were also immunohis-
`tochemically stained with CD31 (BioGenex, San
`Ramon, CA) and CD34 (Signet Laboratories, Ded-
`ham, MA), antibodies highly sensitive for endothe-
`lium. The number of vessels per x20 high-power
`field was counted in the entire bone marrow core
`(each field representing an area of 0.74 mm2), and
`the median was calculated. The area with the highest
`microvessel count in each bone marrow core was
`termed the "hot spot." The microvessel counts of
`control marrows using factor VIII antibody, CD31, or
`CD34 were similar.
`For the three-dimensional reconstruction of the
`normal and leukemic bone marrow, serial 4-,um-thick
`sections were cut and stained with factor VIII-related
`antigen. Representative areas of 500 x 400 ,tm were
`identified on 21 and 29 consecutive sections from
`the normal and leukemic marrow, respectively.
`These specific areas were photographed at X250
`using 35-mm film and a Zeiss (Oberkochen, Ger-
`many) Axiophot light microscope; serial 8 x 12-inch
`prints were made. The photographs were positioned
`under a video camera, and each image was cap-
`tured by a Vision 8 frame grabber(InSync Technolo-
`gies, Inc., Oakland, CA), which was interfaced with a
`personal computer. Images of adjacent sections
`were microaligned while switching between the
`stored image and the live image to minimize motion
`of structures in the selected area.4 6 Using the Vi-
`sion 8 system of tracing programs developed at the
`Children's Hospital Image Graphic Laboratory, the
`stained vessel contours as well as the easily visual-
`ized borders of trabecular bone were outlined on
`each digitized image. These contours, along with an
`assumed 4-,um thickness between slice surfaces,
`were used as input to the Interactive Computerized
`Analysis Reconstruction system program (ISG Tech-
`nologies, Mississauga, Ontario, Canada)7 to gener-
`ate a three-dimensional construct. Nonoverlapping
`serial contours were left disconnected so that no
`data were added to the histological information. The
`Interactive Computerized Analysis Reconstruction
`system was used to display, rotate, and photograph
`the reconstructed images.
`For the quantitative determination of human bFGF
`in the urine we used the Quantikine FGF immunoas-
`say (R & D systems, Inc., Minneapolis, MN), a solid-
`phase enzyme-linked immunosorbent assay method
`that gives a rapid, sensitive, and specific measure-
`ment of bFGF in the urine or any other biological
`fluid.8 This assay uses a quantitative "sandwich" en-
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 002
`
`

`

`817
`Tumor Angiogenesis in Acute Lymphoblastic Leukemia
`AJP March 1997, Vol. 150, No. 3
`
`Table 1.
`
`Vessel Count on Day 0
`
`0.74
`0.28
`
`Nonrelapsed
`(n = 9)
`
`38
`48
`
`Relapsed
`(n = 18)
`40
`49
`
`p
`sO.0001
`<0.0001
`
`Controls
`(n= 10)
`
`68
`
`Leukemia
`(n = 40)
`42
`51
`
`Vessel count
`
`Median
`Hot spot
`
`*Wilcoxon sign rank test.
`
`zyme immunoassay technique, using antibodies
`specific for bFGF and a chromogen and measuring
`the color intensity by optical density.
`
`Results
`The bone marrow microvessel density was markedly
`and significantly increased in children with acute
`lymphoblastic leukemia compared with bone marrow
`controls (Figure 1). The median and highest (hot
`spot) microvessel counts at day 0 (biopsy performed
`at presentation for diagnosis) were markedly high in
`all biopsies with leukemia. The median and highest
`microvessel counts in children with leukemia were 42
`and 51, respectively, versus 6 and 8, respectively, in
`controls (P < 0.0001; Table 1). The numerous mi-
`crovessels seen in leukemic marrow were tortuous,
`often without visible lumina, and appeared packed
`with
`leukemic
`lymphoblasts.
`Small
`endothelial
`"tsprouts" without discernible lumina were frequently
`present, which probably represented newly formed
`microvessels.
`Although the median and highest microvessel
`counts at day 0 bone marrow biopsies were slightly
`lower in the specimens from patients who never re-
`lapsed, compared with those who did relapse, this
`difference was not statistically significant, 38 and 48,
`respectively, in nonrelapsed leukemias versus 40
`and 49, respectively, in leukemias that relapsed (P <
`0.74 and 0.28, respectively; Table 1). The microves-
`sel counts did not change significantly during che-
`motherapy in children in whom serial biopsies on
`days 0, 3, and 31 were performed. Between days 0
`and 3, there was no difference at all in microvessel
`density, with a hot spot median difference of 0 (P <
`0.97). Between days 0 and 31 of therapy there was a
`nonsignificant decrease in microvessel density, with
`a hot spot median difference of -10 (P - 0.6; Table
`2). Although a total of 10 children had biopsies on
`day 31, only for 3 of these children were day 0
`biopsies also available for comparison of microves-
`sel counts. For the other 7 children the day 0 biop-
`sies were either not done (diagnosis made in the
`aspirate) or not suitable for immunohistochemical
`stains. In these 7 children, however, the median
`microvessel count and the hot spot (67 and 80,
`
`respectively) were similar to those of the other 3
`patients. Although numerically the microvessels did
`not change significantly during chemotherapy, there
`was a morphological change that was observed
`early during treatment (from day 3). Microvessels
`became open, empty, and focally dilated, acquiring
`1C).
`(Figure
`sinusoidal
`These
`appearance
`a
`changes were the result of unpacking of tumor cells
`and decreasing tumor burden in the marrow from the
`cytolytic chemotherapy.
`The computer-aided three-dimensional
`recon-
`struction model revealed a complex arborizing
`branching of microvessels in bone marrow with leu-
`kemia, compared with few simple, straight microves-
`sels without branching in controls (Figure 2). The
`shape, size, and caliber of microvessels in leukemia
`were variable and nonuniform, with focal areas of
`narrowings and dilatations, areas of complex loop-
`ings, and formation of microsaccules or diverticuli.
`Urine levels of bFGF at day 0 were markedly high
`in children with leukemia compared with controls.
`The median value in 22 leukemic children was 9075
`pg/g creatinine, compared with 1210 pg/g creatinine
`in 39 age-matched control children (P < 0.0001;
`Table 3).
`
`Discussion
`Previous studies of normal bone marrow have dem-
`onstrated a close association between the bone mar-
`row vessels and the islands of developing blood
`cells, suggesting an important interdependence be-
`tween the bone marrow hematopoietic function and
`its vascular bed.9'10 Indeed, the circulation within the
`bone marrow has been compared with that of major
`organs, such as the spleen or liver.10 Using angio-
`graphic methods in a variety of experimental animal
`models, it has been shown that the nutrient artery
`and vein penetrate the bone together. The artery
`then coils helically around the venous stem, which
`runs along the central axis of the bone. The arterial
`vessels arborize into arterioles and capillaries, which
`open conically into widening sinusoids. These sinu-
`soids are elongated fusiform structures forming a
`complex anastomosing hexagonal network. Direct
`shunting between capillaries and venules permits
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 003
`
`

`

`Perez-Atayde et al
`818
`A/P March 1997, Vol. 150, No. 3
`
`Table 2.
`
`Ljfrcts of Chemnotherapy on Vessel Count
`
`Treatment days
`
`0 and 3 (n = 15)
`0 and 31 (n = 6)
`
`Hot spot
`median difference
`
`0
`-10
`
`P*
`
`0.97
`0.6
`
`*Wilcoxon matched-pairs sign rank test.
`
`the blood to bypass this sinusoidal system. Studies
`in animals have shown the existence of an extensive
`network of capillaries encircling fat cells. Many of
`these capillaries are nonpatent and functionally dor-
`mant. When the activity of the marrow increases,
`some of
`these
`capillaries become
`collapsed
`patent. 1 0
`In the normal bone marrow specimens used as
`controls for this study, the capillaries and sinusoidal
`network were difficult to visualize in routinely stained
`hematoxylin and eosin or Giemsa preparations. With
`the aid of immunohistochemical stains using factor
`VIII-related antigen, CD31, or CD34, the microvessel
`density appeared to be very low (Figure 1A) in com-
`parison with that reported in angiographic or injec-
`tion studies.9'10 This sparse microvessel density may
`be related to the nonfunctionality or dormancy of the
`endothelium in most capillaries and sinusoids or per-
`haps to nonreactivity to these antibodies in normal
`bone marrow capillaries and sinusoids. Nonreactivity
`to factor VIII-related antigen has been reported in
`normal hepatic sinusoids.11 1-15 Hepatic sinusoids be-
`come reactive to factor VIII, acquiring a capillary-
`type basal lamina (capillarization) in certain disor-
`ders of the liver, including chronic hepatitis, alcohol
`liver disease, and nodular regeneration.11-15
`Both routine immunohistochemical studies and a
`computer-generated three-dimensional reconstruc-
`tion model revealed a remarkable increase in the
`number of visualized vessels in bone marrow re-
`placed with leukemia cells (Figures 1 and 2). This
`marked increase in vessel counts might be related to
`reactivation of marrow sinusoids for factor VIII, as
`occurs in the hepatic sinusoid, or alternatively to a
`true neoangiogenic phenomenon. The highly vari-
`able morphology of the microvessels, however, with
`arborizing branching, the presence of endothelial
`sprouts without discernible lumina, the presence of
`hot spots, and the fact that microvessels were ex-
`ceedingly numerous, all suggest a truly neoangio-
`genic phenomenon.
`The microvessel counts at diagnosis remained
`high in biopsies taken after 3 days of single-agent
`antileukemia treatment when tumor cells were mark-
`edly decreased. At this time microvessels appeared
`dilated and empty (Figure 1C). There was, however,
`
`a mild,
`statistically
`insignificant decrease in mi-
`crovessel counts after 1 month of chemotherapy
`when the leukemia was in complete remission. The
`involution of microvessels thus lagged behind the
`killing of leukemia cells. The time necessary for the
`microvessels to regress to a normal number after the
`child is cured of the leukemia remains to be known.
`No biopsies were performed after the first month of
`treatment in children with favorable outcomes.
`
`Figure 1. A: Factor- VII-stained conitrol bonie marrow, shooinlgq venr
`rare capillaries. The enidotheliom qf an arteriole is highlighted in the
`Occasioncal megakarvocyte appear strong/i'
`upper right corner.
`stained. B: Lenkemic bonie miiarrowc! at dayv 0 shooing replacemenelt qi'
`heniatopoietic cells bhy a nionotonouis neoplastic cell popullationi and
`numnerous microvessels highlighted in red by.fictor VIII immmnohisto-
`chefnical staini. Some capillaries are sniiall itithout lmneni, probably
`representing netolyfonned endothelial spvouts. C: Leinkemiiic bone nliar-
`ron' after treatment (day 3) shooing marked decrease in tltum1or cell
`load. Microtvessels, still numierous, appear opetn oith a sinuisoidal
`appearance.
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 004
`
`

`

`Tumor Angiogenesis in Acute Lymphoblastic Leukemia
`819
`A/P March 1997, Vol. 150, No. 3
`
`Figure 2. A: Comnputer-aided three-dimensional reconstnrctionl of control bone narrow' after iimlmnohistochemical staininig for factor VIII. Feu
`stOaClight mnicrovessels (nired) are seen surrounding a firagment of trabecular bonie. B: In conitrast, the three-dimencsional mocdel in this leukemnic
`marrow reveals numerous arborizing and branChing mnicrovessels nubitb irregular contours, withiOrination Of saccular dilations and dietveicIli. The
`areas clepictecd in (A) and (B) hove identical dlimnensionis representing simiiilca amounts ql/mar(ro1'.
`
`The prognostic significance of neovascularization
`in solid neoplasms was first shown in cutaneous
`melanoma. Two reports correlated increased angio-
`genesis with a higher rate of metastases in thin and
`intermediate thickness melanomas.16 17 This conten-
`tion, however, has been recently disputed."8 The
`biological phenomenon of tumor angiogenesis as an
`indication of higher malignant potential, however,
`has been shown on multiple occasions with other
`solid neoplasms.1"19'20 In invasive breast carcinoma,
`microvessel density in the area of most intense an-
`giogenesis was found to be an independent and
`highly significant prognostic indicator of survival
`even in patients without evidence of lymph node
`metastases.19
`
`Table 3.
`
`Urtinie levels of bFGF (pg/l
`
`Creatiniine)
`
`In our study there was no statistically significant
`difference in microvessel density at day 0 between
`relapsed and nonrelapsed leukemias. This lack of
`correlation may be related to the current high sur-
`vival rate of children with acute lymphoblastic leuke-
`mia. This is, however, a preliminary study with a
`small sample of cases. Prospective studies should
`be performed to further clarify this issue.
`Urine bFGF levels at day 0 were markedly in-
`creased in all children with acute lymphoblastic leu-
`kemia, and there was a mild decrease in levels dur-
`ing the first month of treatment. It has been shown
`that bFGF is chemotactic and mitogenic for endothe-
`lial cells in vitro.2122 It also induces breakdown of
`basement membrane proteins and increases colla-
`
`Median*
`Median differencet
`
`Leukemia
`(n = 22)
`9075 pg/g
`NA
`
`Controls
`(n = 39)
`1210 pg/g
`NA
`
`p
`O0.0001
`NA
`
`Effects of chemotherapy
`Days 0 and 31
`P
`NA
`NA
`-3354 pg/g
`0.46
`
`NA, not applicable.
`*Wilcoxon sign rank test.
`tWilcoxon matched-pairs sign rank test.
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 005
`
`

`

`Perez-Atayde et al
`820
`AJP March 1997, Vol. 150, No. 3
`
`genase activity.23 bFGF plays an important role in
`wound repair, neuronal function, and angiogene-
`sis.23 Excessive amounts of bFGF are indicators of
`pathological cell proliferation, such as tumor growth,
`or of neovascularization.2426 It is, therefore, not sur-
`prising that children with acute lymphoblastic leuke-
`mia had markedly elevated urine bFGF levels com-
`pared with age-matched controls (Table 3), and that
`during the first month of chemotherapy the bFGF
`levels decreased concomitantly with decreased tu-
`mor burden. These findings do not indicate the
`source of the increased levels of bFGF in these
`children. It is possible that bFGF is released mainly
`by leukemia cells, by the endothelium, by both tumor
`cells and the endothelium, or perhaps by some other
`source.27
`In mice bearing solid tumors, antiangiogenic ther-
`apy plus conventional cytotoxic therapy causes
`long-term "cures," whereas treatment with either
`therapy alone only inhibits tumor growth.28 These
`data suggest that combination therapy that attacks
`both on the endothelial cell compartment and on the
`tumor cell compartment of a solid tumor may be
`more effective than either therapy alone. It remains to
`be seen whether this two-compartment concept ap-
`plies to leukemia.
`In summary, we have demonstrated that bone
`marrow in childhood acute lymphoblastic leukemia is
`density,
`associated
`increased
`microvessel
`with
`which persists at the time of complete remission.
`Urine bFGF is also elevated at the time of diagnosis
`and decreases with treatment. The pathophysiology
`of these observations suggests that the combination
`of antiangiogenic and cytolytic therapy might apply
`to leukemias, as it does to animal solid tumors; how-
`ever, this remains to be determined.
`
`Acknowledgments
`We are indebted to Dr. Kristen Harris, director of the
`Children's Hospital Image Graphic Laboratory, for
`her assistance in the computer-aided reconstruction
`model and to James A. Edwards III for technical
`assistance.
`
`References
`
`1. Weidner N, Folkman J: Tumoral vascularity as a prog-
`nostic factor in cancer. Important Advances in Oncol-
`ogy. Edited by VT DeVita, S Hellman, SA Rosenberg.
`Philadelphia, Lippincott-Raven, 1997 (in press)
`2. Folkman J: Tumor angiogenesis. The Molecular Basis
`of Cancer. Edited by J Mendelsohn, MJ Howley, MA
`
`Israel, LA Liotta. Philadelphia, WB Saunders, 1995, pp
`206-232
`3. Folkman J, Shing Y: Angiogenesis. J Biol Chem 1992,
`267:10931-10934
`4. Stevens JK, Trogadis J: Computer assisted reconstruc-
`tion from serial electron micrographs. Annu Rev Neu-
`robiol 1984, 5:341-369
`5. Harris KM, Stevens JK: Dendritic spines of rat cerebel-
`lar Purkinje cells: serial electron microscopy with refer-
`ence to their biophysical characteristics. J Neurosci
`1988, 8:4455-4469
`6. Harris KM, Stevens JK: Dendritic spines of CAl pyra-
`midal cells in the rat hippocampus: serial electron mi-
`croscopy with reference to their biophysical character-
`istics. J Neurosci 1989, 9:2982-2997
`7. Pearlstein RA, Kirschner L, Simons J, Machell S, White
`WF, Sidman RL: A multimodal system for reconstruc-
`tion and quantification of neurologic structures. Anal
`Quant Cytol Histol 1986, 8:108-115
`8. Nguyen M, Watanabe H, Budson AE, Richie JP, Hayes
`DF, Folkman J: Elevated levels of an angiogenic pep-
`tide, basic fibroblast growth factor, in the urine of pa-
`tients with a wide spectrum of cancers. J Natl Cancer
`Inst 1994, 86:356-361
`9. Bruyn PPH, Breen PC, Thomas TB: The microcircula-
`tion of the bone marrow. Anat Rec 1970, 168:55-68
`10. Branemark P-I: Bone marrow microvascular structure
`and function. Adv Microcirc 1968, 1:1-65
`11. Bhunchet E, Fujieda K: Capillarization and venulariza-
`tion of hepatic sinusoids in porcine serum-induced rat
`liver fibrosis: a mechanism to maintain liver blood flow.
`Hepatology 1993, 18:1450-1458
`12. Martinez-Hernandez A, Martinez J: The role of capillar-
`ization in hepatic failure: studies in carbon tetrachlo-
`ride-induced cirrhosis. Hepatology 1991, 14:864-874
`13. Dubuisson L, Boussarie L, Bedin C-A, Balabaud C,
`Bioulac-Sage P: Transformation of sinusoids into cap-
`illaries in a rat model of selenium- induced nodular
`regenerative hyperplasia: an immunolight and immu-
`noelectron microscopic study. Hepatology 1995, 21:
`805-814
`14. Garcia-Monzon C, Sanchez-Madrid F, Garcia-Buey L,
`Garcia-Arroyo A, Garcia-Sanchez A, Moreno-Otero R:
`Vascular adhesion molecule expression in viral chronic
`hepatitis: evidence of neoangiogenesis in portal tracts.
`Gastroenterology 1995, 108:231-241
`15. Urashima S, Tsutsumi M, Nakase K, Wang J-S, Takada
`A: Studies on capillarization of the hepatic sinusoids in
`alcoholic liver disease. Alcohol Alcohol 1993, 28:77-84
`16. Graham C, Rivers J, Kerbel R, Stankiewicz K, White W:
`Extent of vascularization as a prognostic indicator in
`thin (<0.76 mm) malignant melanomas. Am J Pathol
`1994, 145:510-514
`17. Srivastava A, Laidler P, Davies R, Horgan K, Hughes L:
`The prognostic significance of tumor vascularity in in-
`(0.76-4.0 mm thick)
`skin
`termediate-thickness
`melanoma: a quantitative histologic study. Am J Pathol
`1988, 133:419-423
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 006
`
`

`

`821
`Tumor Angiogenesis in Acute Lymphoblastic Leukemia
`AJP March 1997, Vol. 150, No. 3
`
`18. Busam KJ, Berwick M, Blessing K, Fandrey K, Kang S,
`Karaoli T, Fine J, Cochran AJ, White WL, Rivers J, Elder
`DE, Wen D-RP, Heyman BH, Barnhill RL: Tumor vas-
`cularity is not a prognostic factor for malignant mela-
`noma of the skin. Am J Pathol 1995, 147:1049-1056
`19. Weidner N, Semple JP, Welch WR, Folkman J: Tumor
`angiogenesis and metastases- correlation in invasive
`breast carcinoma. N EngI J Med 1991, 324:1-8
`20. Maeda K, Chung Y-S, Takatsuka S, Ogawa Y, Sawada
`T, Yamashita Y, Onoda N, Kato Y, Nitta A, Arimoto Y,
`Kondo Y, Sowa M: Tumor angiogenesis as a predictor
`of recurrence in gastric carcinoma. J Clin Oncol 1995,
`13:477-481
`21. Chodak GW, Scheiner CJ, Zetter BR: Urine from pa-
`tients with transitional-cell carcinoma stimulates migra-
`tion of capillary endothelial cells. N Engl J Med 1981,
`305:869-874
`22. Baird A, Klagsbrun M: The fibroblast growth factor
`family, an overview. The Fibroblast Growth Factor Fam-
`ily, vol 638. Edited by A Baird, M Klagsbrun. New York,
`New York Academy of Sciences, 1991, pp xi-xii
`23. Rifkin D, Gross J, Moscatelli D, Jaffe E: Proteases and
`angiogenesis: production of plasminogen activator
`and collagenase by endothelial cells. Pathobiology of
`
`the Endothelial Cell. Edited by H Nossel, H Vogel. New
`York, Academic Press, 1982, pp 191-197
`24. Chodak GW, Hospelhorn V, Judge SM, Mayforth R,
`Koepper H, Sasse J: Increased levels of fibroblast
`growth factor-like activity in urine from patients with
`bladder or kidney cancer. Cancer Res 1988, 48:2083-
`2088
`25. Fujimoto K, Ichimori Y, Kakizoe T, Okajima E, Sakamoto
`H, Sugimura T, Terada M: Increased serum levels of
`basic fibroblast growth factor in patients with renal cell
`carcinoma. Biochem Biophys Res Commun 1991, 180:
`386-392
`26. Nguyen M, Watanabe H, Budson AE, Richie JP, Folk-
`man J: Elevated levels of the angiogenic peptide basic
`fibroblast growth factor in urine of bladder cancer pa-
`tients. J Natl Cancer Inst 1993, 85:241-242
`27. Brunner G, Nguyen H, Gabrilove J, Rifkin DB, Wilson
`EL: Basic fibroblast growth factor expression in human
`bone marrow and peripheral blood cells. Blood 1993,
`81:631-638
`28. Teicher BA, Holden SA, Ara G, Sotomayor EA, Huang
`ZD, Chen Y-N, Brem H: Potentiation of cytotoxic cancer
`therapies by TNP-470 alone and with other anti-angio-
`genic agents. Int J Cancer 1994, 57:920-925
`
`Breckenridge Exhibit 1041
`Perez-Atayde
`Page 007
`
`