Uterine Cervical Carcinoma: Comparison of Standard and Pharmacokinetic Analysis of Time-Intensity Curves for Assessment of Tumor Angiogenesis and

Patient Survival
Hans Hawighorst, Paul G. Knapstein, Michael V. Knopp, et al. Cancer Res 1998;58:3598-3602.

Updated version

Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/58/16/3598

E-mail alerts Reprints and Subscriptions Permissions

Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at pubs@aacr.org. To request permission to re-use all or part of this article, contact the AACR Publications Department at permissions@aacr.org.

Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.

[CANCER RESEARCH 58, 3598-3602.

August 15. 1998|

Uterine Cervical Carcinoma: Comparison of Standard and Pharmacokinetic Analysis of Time-Intensity Curves for Assessment of Tumor Angiogenesis and Patient Survival
Hans Hawighorst,1 Paul G. Knapstein, Michael V. Knopp, Wolfgang Weikel, Gunnar Brix, Ivan Zuna, Stefan O. Schnberg, Marco Essig, Peter Vaupel, and Gerhard van Kaick
Department of Radiological Diagnostics. German Cancer Research Center, D-69120 Heidelherg \H. H.. M. V. K., C. B., I. Z., S. O. S., M. ., G. v. K.I. and Department Obstetrics and G\necolog\ IP. G. K.. W. W.J and Institute of Ph\'siolog\ and Paihophysiology P.V.l. University of Mainz. D-55101 Mainz, Germany of

ABSTRACT
Dynamic studies of Gd-based contrast agents in magnetic resonance imaging (MRI) are increasingly being used for tumor characterization as well as for therapy response monitoring. Because detailed knowledge regarding the pathophysiological properties, which in turn are responsible for differences in contrast enhancement, remains fairly undetermined, it was the aim of this study to: (a) examine the association of standard and pharmacokinetic analysis of time-intensity curves in dynamic MRI with histomorphological markers of tumor angiogenesis [microvessel density (MVD) and vascular endothelial growth factor (VEGF)]; and (In deter mine the ultimate value of a histomorphological and a dynamic MRI approach by the correlation of those data with disease outcome in patients with primary cancer of the uterine cervix. Pharmacokinetic parameters (amplitude, A; exchange rate constant, *2I) and standard parameters [the maximum signal intensity increase over baseline iSI-I i and the steepest signal intensity-upslope per second |SI-17 s)] were calculated from a contrast-enhanced dynamic MRI series in 37 patients with biopsy-proven primary cervical cancer. On the surgical whole mount specimens, histomorphological markers of tumor angiogen esis (MVD and VEGF) were compared to MRI-derived parameters. For MRI and histomorphological data, Kaplan-Meier survival curves were calculated and compared using log-rank statistics. A significant association was found between MVD and A (/' < 0.01) and SI-I (P < 0.05). No significant relationships were observed between VEGF expression and all dynamic MRI parameters. Kaplan-Meier curves based on 21 and SI-U/s showed that tumors with high *_,, and SI-U/s values had a significantly (P < 0.05 and 0.001, respectively) worse disease outcome than did tumors with low A,, and SI-U/s values. None of the histomor phological gold standard markers for assessing tumor angiogenesis (MVD and VEGF) had any significant power to predict patient survival. It is concluded that in patients with uterine cervical cancer: (a) the pathophysiological basis for differences in dynamic MRI is MVD but not VEGF expression; (In a functional, dynamic MRI approach (both stand ard and pharmacokinetic analysis) may be better suited to assess angiogenic activity in terms of patient survival than are the current histomorphological-based markers of tumor angiogenesis; and (< i compared with standard analysis, a simple pharmacokinetic analysis of time-intensity curves is not superior to assess MVD or patient survival.

tumor cells into the systemic circulation and hence the formation of distant mtastases(1). Although the histological MVD2 technique is the current gold standard to characterize tumor angiogenesis, it may not be an ideal tool for clinical purposes, because it is invasive, is prone to sampling errors, and does not assess the functional angiogenic activity (1,6). In addition, there is strong evidence that angiogenic activity is also reflected in the vascular permeability of tumors. VEGF was identified independently as a powerful enhancer of endothelial hyperpermeability (7), and the leakage of serum proteins is considered an essential step in the angiogenic process (8). Clearly, an in vivo assay for estimating angiogenic activity would be desirable and clinically pow erful, particularly if the method were quantitative, noninvasive, able to sample the entire tumor, and could be repeated at frequent intervals. Dynamic contrast-enhanced MRI is gaining acceptance as a pow erful new tool to assess the spatial and temporal heterogeneity of tumor microcirculation (9-11). Based on the time-intensity curves, different standard and pharmacokinetic MR postprocessing analyses have been proposed (9-15). For the standard evaluation, previously described parameters were estimated (Refs. 9, 10, 14, and 15; SI-I and SI-U/s). For the pharmacokinetic analysis, A and k2l were computed using a simple pharmacokinetic model (13). However, despite the enormous clinical potential to assess func tional tumor angiogenesis, detailed MRI correlations with present gold standards of tumor angiogenesis (MVD and VEGF expression) in a clinical setting are not available. One major obstacle is that accurate correlations necessitate anatomical MRI maps to be com pared with surgical whole mount specimens and not unrepresentative bioptic specimens, so that anatomically similar-positioned ROIs can be set. The purpose of this study is to: (a) examine the correlation of standard and pharmacokinetic analysis of time-intensity curves in dynamic MRI with histomorphological markers of tumor angiogene sis (MVD and VEGF); and (b) determine the ultimate value of a histomorphological and a dynamic MRI approach by correlation of those data with disease outcome in patients with primary cancer of the uterine cervix.

INTRODUCTION Neovascularization is a fundamental prerequisite for expansive and clinically relevant growth of solid tumors ( 1). Various experimental models as well as clinicopathological observations have shown that solid tumors [e.g., breast (2), lung (3), and cervical carcinoma (4, 5)] cannot attain diameters larger than 2 mm without their own vascular supply. Moreover, new vessels are obligatory for the shedding of
Received 2/27/98; accepled 6/16/98. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely lo indicate this fact. 1To whom requests for reprints should be addressed, at Department of Radiological Diagnostics and Therapy. German Cancer Research Center. Im Neuenheimer Feld 280. D-69120 Heidelberg. Germany. Phone: 49-6221-422525; Fax: 49-6221-422531: E-mail: H.Hawighorst@dkfz-heidelberg.de.

MATERIALS
Patients.

AND METHODS
patients (mean age, 49 14 years) with biopsy-

Thirty-seven

proven primary cancer of the uterine cervix (International Federation of Gynecologists and Obstetricians stage: IB, 6 patients; HA, 1 patient; IIB. 20 patients; IIIB, 4 patients; and IVA, 6 patients) were included in this study. Surgical treatment consisted of radical hysterectomy (n = 27) or pelvic exenteration (= 10) with pelvic lymph node dissection in all cases.
2 The abbreviations used are: MVD. microvessel density: MRI. magnetic resonance

imaging: MR. magnetic resonance; RO,region of interest; VEGF, vascular endothelial growth factor; SI-I. signal intensity increase over baseline: Sl-U/s. the steepest signal inlensity-upslope per second; A, maximal amplitude of enhancement; &21.tissue exchange rate constant; TR. repetition time; TE, echo time: Gd-DTPA. gadolinium megluminpentetate.

3598

Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.

ASSESSMENT

OF TUMOR ANGIOGENESIS

BY DYNAMIC

MRI

MRI Protocol.

All MR studies were performed on a 1.5-Tesla clinical MR

system (Magnetom SP 4000 or Vision; Siemens, Erlangen, Germany) using a pelvic surface or phased-array coil. The MR examination included T2weighted fast spin-echo images (TR/TE = 2800-4200/93-120 ms; field of view = 180-350 mm; section thickness = 5-7 mm; 3-15 echoes) as well as pre- and postcontrast T1-weighted spin-echo images (TRA"E = 600/15 ms; field of view = 180-350 mm; section thickness = 5-7 mm). A strict dynamic MRI protocol was adhered to to minimize errors derived from the MRI data acquisition. The following points were taken into account: (a) the entire cervical tumor was scanned by 8-10 sections with 22 repetitions each; (b) an ultrafast and linearly T l-weighted saturation recovery, turbo fast low-angle shot (TR/TE = 10/4 ms; flip angle a = 12 degrees; recovery time = 200 ms; matrix = 256 X 128 interpolated to 256 X 256) sequence with a high image repetition ratio of 1.4 s/section was used (12); (c) the effect of operator dependency on injection times was reduced by a short constant rate infusion within 30 s using a variable-speed infusion pump (CAI 626P; Doltron AG, Uster, Switzerland); (d) the lag-time between the injection of contrast material and the individual arrival at the tumor was estimated from the time-intensity course measured in the iliac arteries; (e) quantitative data analysis was performed at a high in-plane resolution (0.9-1.4 X 0.9-1.4 mm) of a dynamic time course; and (/) to accurately compare similar-sized and -positioned ROIs on the whole mount specimens and on the corresponding MRI-derived maps, all histopathological/MRI correlations were performed by matching whole mount specimens with MRI sections cut in an identical medioaxial or mediosagittal plane. Data Analysis. The paramagnetic contrast material Gd-DTPA (Magnevist, Schering, Berlin) was used at a total dose of 0.1 mmol Gd-DTPA/kg body weight. The automated processing of the acquired dynamic MRI data sets on a pixel-by-pixel basis for each section was performed on a DEC-3000 work station (model 500; Digital Equipment Co., Maynard, MA). In addition, the calculated MR images were color-encoded to improve the visibility of tumor regions with different contrast enhancement (12). Based on the time-intensity curves, different MR postprocessing analyses were performed. For the stand ard evaluation, previously described parameters were estimated [Refs. 9, 10, 14, 15; SI-I and SI-U/s (Fig. 1)]. For the pharmacokinetic analysis, A and k2l were computed using a simple pharmacokinetic model (13). MRI-based Histopathological Correlation. All surgical specimens were cut in a mediosagittal or medioaxial plane resembling (with an approximate accuracy of 3-5 mm) the sections obtained by dynamic MRI. The surgical specimens were processed by the macroslide technique to verify histopathological tumor extension. The whole mount sections of 5 /m were stained with immunohistochemical factor Vlll-related antigen. Criteria for vessel counting were those established by Weidner et al. (2), and each ROI was counted twice by two independent pathologists. The mean counts (marked with 2 X 2-mm rectangles) with the highest MVD of at least 2-4 hot spots were recorded and

identified at a X100 magnification (16). In the present study, corresponding tissue sections and color-coded MRI maps were evaluated. Thereafter, the mean values of at least two to four ROIs obtained on the MRI color-coded maps encoding the maximal MRI parameters (A, k2, SI-I, and SI-U/s) were recorded and compared with the histomorphological markers. Staining for VEGF, a tumor marker for vascular permeability and angiogenesis, was performed using the commercially available polyclonal antiVEGF antibody AB-2. The mean counts with the highest anti-VEGF staining of 10 hot spots were recorded and identified at a X400 magnification. For the evaluation of VEGF expression, immunoreactivities were graded on a fourpoint scale (scored as positive around the vessels and stromal staining; 1 = none; 2 = slight; 3 = moderate; and 4 = strong) according to the staining intensity around vessels including the stromal intensity (17). Statistical Analysis. Statistical analysis of the histopathological and dy namic MR data was performed using SAS software (SAS for Windows, Version 6.10; SAS, Cary, NC). To test the strength of association between the two histomorphological markers and the MRI parameters, we used the un paired ic test. All cases were stratified into high and low MR parameters and histomorphological data according to the median value. For both groups, Kaplan-Meier survival curves were calculated and compared using log-rank statistics. P < 0.05 was considered statistically significant.

RESULTS After a median follow-up time of 24 months (range, 5-39 months), 14 of 37 patients (38%) died of relapsed disease. Individual and median MVD, VEGF score, and median dynamic MRI parameters are summarized in Tables 1 and 2. The intra- and interobserver variabilities in MVD measurements were 4 and 7%, respectively. Correlation of similarly positioned ROIs between the maximal histopathological and the MRI data was possible in all tumors. The median MVD in a X 100 field for all tumors was 30 vessels/4 mm2 (range, 7-45 vessels/4 mm2). The intratumoral MVD and VEGF heterogeneity from microregion to microregion was mod erate, and the median MVDs and VEGFs on the macroslide specimens obtained in eight different areas exhibited a significant association (P < 0.05). For individual tumors, two main patterns of MVD expression were noted: (a) low values for central MVD and high values for peripheral MVD were observed, although no significant differences in the mean MVD between the central and peripheral regions could be detected; and (b) a homogeneous distribution of MVD over the whole mount specimen was recorded.

3.0 _|

F. 2.5 -i

2.0 Fig. 1. Time-intensity curve derived from a ROI as a basis for different MRI postprocessing analy ses. For standard evaluation, SI-I and the steepest SI-U/s were used. For the pharmacokinetic analysis, A and k2were computed.

SI-U/s
1.5 -

1.0

II
-1.0
0.0

I 3..0

I 4.0

1.0

2.0

! 5.0

I
6.0

I
7.0

time (min)

3599
Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.

ASSESSMENT

OF TUMOR ANGIOGENESIS

BY DYNAMIC

MR1

Table 1 Individual ofPatient histomorpholngical

findings in 37 patients with primary cancer

DISCUSSION Recent studies have shown that differences in tumor angiogenesis are correlated with different patterns of contrast enhancement ob served by MRI (9-11, 14, 16). Because angiogenesis plays a pivotal role in the pathogenesis of tumor growth and metastasis, and a number of potential therapeutic interventions based on angiogenic phenomena are to be expected in the foreseeable future, a more detailed under standing of the pathophysiological basis responsible for the differ ences in MR contrast enhancement patterns is necessary. On the other hand, a variety of techniques including pharmacokinetic models have been proposed for the analysis of time-intensity curve measurements by dynamic MRI (9-13). However, no studies exist comparing a simple pharmacokinetic analysis with a standard and commonly less time-consuming analysis to estimate angiogenic activity and tumorbiological aggressiveness in terms of patient survival. Therefore, this study was aimed at improving our in vivo understanding of the angiogenic properties responsible for MR contrast enhancement and to determine which MRI data analysis strategy can be implemented in clinical practice to reflect tumor angiogenesis. A T2*-weighted dy namic MRI approach was excluded from this study, because there are no intravascular contrast agents available at present. This study demonstrates that there are widely heterogeneous pat terns of fast and high contrast enhancement in human cervical tumors that have to be accounted for when analyzing dynamic MR images. Areas of high contrast medium uptake do not necessarily coincide with areas of fast contrast medium uptake. A color-coded map pro vides important visual insights into the spatial variations of tumor microcirculation that cannot be appreciated on the gray scale MR images alone. The present work has implications for the selection of ROI-based MRI histomorphological correlations. Areas with the highest contrast medium uptake (A and SI-I) colocalized significantly with focal hot spots of MVD, as estimated on the whole mount specimens. Because the native Tl relaxation time did not show significant differences between malignant tumors, and, moreover, because the MR sequence parameters and Gd-based infusion rates were kept constant, parame ters A and SI-I mainly represent the size of the intravascular blood volume. A further fundamental finding of this study is that areas of high MVD and high microvessel permeability do not necessarily colocalize, as evidenced by the inverse association between MVD and VEGF expression, on the one hand, and large spatial variations of areas with a high and fast MRI contrast uptake, on the other hand. Theoretically, the rate (upslope) of contrast media uptake is de pendent on multiple factors, including the perfusion rate, MVD, microvessel permeability, and the size of the extracellular leakage space (9-11, 14). Which of these pathophysiological mechanisms in human tumors is the major contributor for the differences in the velocity of contrast media uptake is not clear. In a recent study using mice, it has been postulated that differences in tumor microvessel

ceirMVD"102717172333263613293016362024203431152745163235292332198332340363471632VEGF4124213421222413231431113323422234133" uterine
no.123456789HiII121314151171819202122232425262728243031323334353637the

MVD per 4 mrrT.

The color-coded MR maps also exhibited two main patterns of contrast enhancement: (a) there was a homogeneously colored map encoding the entire tumor with a fast (fc2, and SI-U/s) and/or high (A and SI-I) contrast medium uptake; and (b) there was a spatial distri bution of contrast uptake with a high uptake in the tumor periphery and/or a fust uptake in the tumor center (Fig. 2). However, these spatial differences in the MR parameters did not reach significance. The ROI-based analysis of the standard and pharmacokinetic pa rameters revealed significant associations (P < 0.01) between SI-I and A on the one hand and SI-U/s and &->,(P < 0.05) on the other hand. Positive VEGF expression was observed in 18 of 37 (49%) tumors; negative VEGF expression was observed in the 19 remaining tumors. Areas of high VEGF expression were commonly associated with a low MVD, and those with a low VEGF expression were commonly associated with a high MVD. This inverse association between VEGF (median score, 2; range, 1-4) expression and MVD was statistically significant (P < 0.02). A significant association was found between MVD and A (P < 0.01), SI-I (P < 0.05), and k2l (P < 0.05). No significant associations were found between any of the dynamic MRI parameters (A, k2l, SI-I, and SI-U/s) and VEGF expression (Table 3). Univariate analysis of survival is shown in Fig. 3. When stratified into high and low median standard and pharmacokinetic parameter groups, median SI-U/s and As, values of >0.18/s and 4.0 min"1 were

Table 2 Median values and ranges of the histomorphological and MRI parameters derived from ti dynamic MRI series in 37 patients with cen-ical cancer (range)Histomorphology MVD"VEGFMRIAk Median values

(7^*5)2(1-4)1.6(0.3-2.2)4.0

the only significant (P < 0.001 and 0.05, respectively) factors in ImiiT'lSI-1SI-U/s30 predicting poor patient survival. None of the histopathological mark (0.3-2.8)0.18(0.04-0.61)" ers of angiogenesis (MVD or VEGF expression) showed any predic MVD per 4 mm2 tive power. 3600

(0.7-23) 1.4

Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.

ASSESSMENT

OF TUMOR ANGIOGENESIS

BY DYNAMIC

MR1

Fig. 2. Ccr\ ical carcinoma ut a 4l)-vear-old patient u ilh International Federation ot Gynecologists and Obstetricians stage 11 H disease. The patient died troni poh ic side recurrence 7 months after a radical hysterectomy. A, mediosagittal whole mount specimen (stained with H&E) depicts a large cervical tumor. B, the mediosagittal color-coded MRI slice shows variations in the spatial distribution of areas of high contrast enhancement (yellow encoded areas; SI-I = 2.3; A = 2.0) in the tumor periphery and fast contrast enhancement (pink-blue'. SI-U/s 0.21/s; Jt21 5.0 min"1) of the more central tumor areas. C magnification (X 100; arrow) of one of the hot spots reveals a high MVD (34 vessels/4 mm2), confirming that focal hot spots of MVD colocalize with areas of high signal intensity on dynamic MRI. The significant (P < 0.02) association between low VEGF expression and high MVD is shown in D. Areas of low VEGF expression (tumor and endothelial cells not intensively stained by the antibody against the VEGF receptor; score = 2) were commonly associated with a high MVD. No significant associations were observed between the MRI-based parameters and VEGF expression. However, when stratified inlo high and low median standard and pharmacokinetic parameter groups, high median SI-U/s and k2t values were the only significant factors in predicting poor patient survival.

permeability can be assessed by Gd-DTPA-based contrast agents (11). However, in the present study examining human tumors in a clinical setting, the detailed spatial analysis of the MRI-based maps revealed no significant association between VEGF expression (a recognized marker for tumor angiogenesis, especially microvessel permeability) and areas of fast contrast uptake (SI-U/s and k2,). Because none of the

Table 3 Association benveen hisluptllhologiciil markers of angiogenesis and dynamic MRl-based parameters

k-,1 (min" ') SI-ISI-U/sMVD per 4 mm2. a bParameterA NS. not significant.MVD"/><0.01

P < 0.05 P < 0.05NSVEGFNS''NS

NSNS

histomorphological markers of tumor angiogenesis showed any sig nificant association with the initial steepest upslope of contrast media uptake (SI-U/s). it may be speculated that SI-U/s is mainly dominated by differences in tissue perfusion. In addition, the pharmacokinetic parameter &-,,is influenced by MVD and tissue perfusion. The findings that VEGF expression did not colocalize with standard and pharmacokinetic MRI parameters are not necessarily contradic tory to the above-mentioned statement that dynamic MRI can estimate tumor permeability. The results of this study substantiate recent find ings that the overall angiogenic activity of tumor cells is adjustable and is continuously modulated according to the metabolic demands (18. 19). It has been shown that once the increased nutritional supply is matched by an efficient increase in MVD and permeability. VEGF expression is down-regulated, specifically in regions already invaded by microvessels (18-20). Apparently, tumors may be in a dynamic

3601
Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.

ASSESSMENT OF TUMOR ANO1OGENESIS BY DYNAMIC

MR1

1.0

SI-U/s<.18 n = 18
-

0.8
D. I

0.6

0.4

0.2

SI-U/s>.18 n = 19 Log-rank P < .001 10

20 30 40 Time (months)

0.0

possible predictive value for the maximal SI-U/s should be seen as a preliminary finding given the small number of patients in the present study. Therefore, additional studies with larger numbers of patients are needed to verify these promising results. The pharmacokinetic analysis also predicted poor patient survival: however, a lower level of signifi cance was reached as compared with the standard analysis. We conclude from our results that in uterine cervical cancer: (a) the pathophysiological basis for differences in dynamic MRI is MVD but not VEGF expression; (b) a functional, dynamic MRI approach (both standard analysis and a pharmacokinetic analysis) may be better suited to assess angiogenic activity in terms of patient survival than are the current histomorphological-based markers of tumor angiogen esis; and (c) compared with standard analysis, a simple pharmacoki netic analysis of time-intensity curves is not superior for the assess ment of MVD or patient survival.

k21< 4 min'1

REFERENCES
1. Folkman. J. Intratumoral microvascular density as a prognoslic factor in cancer. Am. J. Pathol.. 147: 9-19, 1995. 2. Weidner, N.. Joseph, P. S.. William. R. W., and Folkman, J. Tumor angiogenesis and mtastases: correlation in invasive breast carcinoma. N. Engl. J. Med.. 324: 1-8, 1991. 3. Macciarini, P.. Fontanini. G.. Hardin. M. J.. Squartini, F.. and Angeletti. C. A. Relation of neovascularization to metastasis of non-small-cell lung cancer. Lancet, 340: 145-146, 1992. 4. Wiggins, D. L., Granai. C. O., Steinhoff, M. M., and Calabresi, P. Tumor angiogen esis as a prognostic factor in cervical carcinoma. Gynecol. Oncol.. 56: 353-356, 1995. 5. Schlenger. K., Hoeckel, M., Mitze, M.. Schaeffer, U.. Weikel, W., Knapstein, P. G., and Lamben. A. Tumor vascularity: a novel prognostic factor in advanced cervical carcinoma. Gynecol. Oncol., 59.' 57-66. 1995. 6. Bernsen, H. J. J. A., Rijken, P. F. J. W.. Oostendorp, T., and van der Kogel. A. J. Vascularity and perfusin of human gliomas xenografted in the athymic nude mouse. Br. J. Cancer, 71: 721-726, 1995. 7. Brown. L. F.. Berse. B.. Jackman. R. W.. Tognazzi. K.. Guidi. A. J.. Dvorak, H. F., Senger. D. R.. Connolly. J. L., and Schnitt, S. J. Expression of vascular permeability endothelial growth (factor) and its receptors in breast cancer. Hum. Pathol., 26: 86-91, 1995. 8. Kim, K. J., Li, B.. Winer. J., Armanini. M.. Gillett, N., and Phillips. H. S. Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumor growth in vivo. Nature (Lond.). 362: 841-844, 1993. 9. Verstraete. K. L.. De-Deene, Y. D., Roels. H.. Dierick, A., Uyttendaele, D., and Kunnen, M. Benign and malignant musculoskeletal lesions: dynamic contrast-en hanced MR imaging: parametric first-pass images depict tissue vascularization and perfusion. Radiology, 192: 835-843. 1994. 10. Buadu. L. D.. Murakami, J.. Murayama, S.. Hashiguchi. N., Sakai, S.. Masuda. K.. Toyoshima, S., Kuroki, S., and Ohno, S. Breast lesions: correlation of contrast medium enhancement patterns on MR images with histopathologic findings and tumor angiogenesis. Radiology, 200: 639-649, 1996. 11. Degani. H., Gussi, V.. Weinstein, D., Fields, S., and Strano, S. Mapping pathophys iological features of breast tumors by MRI at high spatial resolution. Nat. Med., 3: 780-782, 1997. 12. Hoffmann, U., Brix. G., Knopp. M. V., Hess, T., and Lorenz, W. J. Pharmacokinetic mapping of the breast: a new method for dynamic MR-mammography. Magn. Reson. Med., 33: 506-514. 1995. 13. Brix. G., Semmler, W., Port, R., Schad, L. R., Layer, G., and Lorenz, W. J. Pharmacokinetic parameters in CNS Gd-DTPA enhanced MR imaging. J. Comput. Assisted Tomogr.. 15: 621-627, 1991. 14. Koenig. H., Sieper. J., and Wolf, K. J. Rheumatoid arthritis: evaluation of hypervascular and fibrous pannus with enhanced dynamic MR imaging enhanced with GdDTPA. Radiology, 176: 473-477. 1990. 15. Takashima. S., Noguchi, Y., Okumara. T.. Aruga. H.. and Kobayashi, T. Dynamic MR imaging in the head and neck. Radiology, 189: 813-821, 1993. 16. Hawighorst, H., Knapstein, P. G.. Weikel, W.. Knopp, M. V.. Zuna, I.. Knof, A., Brix, G., Schaeffer. U., Wilkens, C.. Schoenberg. S. O.. Essig, M., Vaupel. P., and van Kaick, G. Angiogenesis of uterine cervical carcinoma: characterization by pharma cokinetic magnetic resonance parameters and histological microvessel density with correlation to lymphatic involvement. Cancer Res.. 57: 4777-4786. 1997. 17. Mattern. J.. Koomgi.R., and Volm, M. Association of vascular endothelial growth factor expression with intratumoral microvessel density and tumor cell proliferation in human epidermoid carcinoma. Br. J. Cancer, 73: 931-934. 1996. 18. Neeman. M.. Ahramovitch. R.. Schiffenbauer. Y. S.. and Tempel, C. Regulation of angiogenesis by hypoxic stress: from solid tumors lo the ovarian follicle. Int. J. Exp. Pathol.. 78: 57-70. 1997. 19. Shweiki. D., Neeman. M.. Itin. A., and Keshet. E. Induction of vascular endothelial growth factor expression by hypoxia and by glucose deficiency in multiple spheroids: implications for tumor angiogenesis. Proc. Nati. Acad. Sci. USA. 92: 768-772. 1995. 20. Stein, I.. Neeman, M., Shweiki. D.. Itin, A., and Keshet, E. Stabilization of vascular endothelial growth factor mRNA by hypoxia and hypoglycemia and coregulation with other ischemia-induced genes. Mol. Cell. Biol.. 15: 5363-5368, 1995.

0.0 20 30 40 Time (months)

1.0 0.8 -

= 22

0.6 -

0.4 -

0.2 -

MVD < 31 n = 15 Log-rank P = .1

0.0

10

20 30 Time (months)

40

Fig. 3. When stratified into high and low median standard and pharmacokinetic parameter groups, median Sl-U/s (A) and 21 (^) values (>0.18 s and 4.0 min" ') were the only significant (P < 0.001 and 0.05. respectively) factors in predicting poor patient survival. None of the histomorphological markers of angiogenesis (MVD or VEGF expression. O showed any predictive power.

steady state in which the hypoxic regions that were recently upregulated will later become the highly metastatic angiogenic hot spots with low VEGF expression. Such tumor dynamics and highly meta static areas cannot be assessed sufficiently by a histomorphological approach, whereas contrast-enhanced dynamic MRI can contribute a lot. Several analytical techniques including simple time-consuming phar macokinetic models have been proposed to evaluate dynamic contrastenhanced MR images (9-13). In the present study, the findings of a
3602

Downloaded from cancerres.aacrjournals.org on July 13, 2013. 1998 American Association for Cancer Research.