Download Antivascular cancer treatments: imaging biomarkers in

The British Journal of Radiology, 76 (2003), S83–S86
DOI: 10.1259/bjr/15255885
E
2003 The British Institute of Radiology
Antivascular cancer treatments: imaging biomarkers in
pharmaceutical drug development
S M GALBRAITH, MB, BChir, PhD
Clinical Discovery, Bristol-Myers Squibb, Princeton, NJ, USA
Over the last several years there have been a large
number of drugs brought into the clinic for the treatment
of cancer targeting antiangiogenic or antivascular mechanisms. The matrix metalloproteinase inhibitors (MMPIs) are
an example of one group of antiangiogenic agents which
were first tested in man beginning 1997 [1]. Pre-clinical
studies had demonstrated inhibition of endothelial cell
growth, inhibition of angiogenesis with matrigel plug
assays, and inhibition of metastasis. There was no
expectation from such studies though that these agents
would cause tumour regression in the manner that
cytotoxic drugs do. Thus, the pattern of development of
oncology agents where in Phase I doses are escalated until
dose limiting toxicity is seen and the next lowest dose
is chosen as the maximum tolerated dose (MTD) was
perceived to be inappropriate [2]. It was proposed that for
drugs with a wider therapeutic window than traditional
cytotoxics, and a mechanism of action that may lead to
cytostasis, rather than tumour regression, that an ‘‘optimal
biological dose’’ (OBD) lower than the MTD might be
more appropriate to take forward into Phase II. However
in order to discriminate the OBD some means of measuring the desired biological activity or ‘‘biomarker’’ is
needed. At the time when the MMPIs entered the clinic
very few such biomarkers were available. However some
exploratory markers were examined by more recent
entrants to the antiangiogenic field such as endostatin
[3, 4], the vascular endothelial growth factor (VEGF)
pathway targeted drugs [5–8] and vascular targeting agents
[11, 12]. In addition to biomarkers in tumour tissue and
blood samples, a range of imaging technologies are
available or under development that have the potential
for use as biomarkers of antiangiogenic and antivascular
drug effects. These can be categorized by the distance
‘‘downstream’’ from the drug interaction with its molecular target:
(1) target inhibition — in vivo MMP2 or 9 inhibition for
example for a MMPI, or inhibition of vascular
endothelial growth factor (VEGF)-R2 phosphorylation for a VEGF-R tyrosine kinase inhibitor.
(2) measurement of an effect on the tumour microvasculature — a change in blood flow, vessel permeability or
blood volume.
(3) measurement of an effect on tumour metabolism,
proliferation or rate of cell death.
If these types of biomarkers are to be used to affect
decisions about drug development such as choice of dose
Address correspondence to Dr Susan M Galbraith, Bristol-Myers
Squibb, Clinical Discovery, Pharmaceutical Research Institute,
PO Box 4000, Princeton NJ 08453-4000, USA.
The British Journal of Radiology, Special Issue 2003
or schedule, or to stop further development of the drug
in the absence of a measurable change in the biological
endpoint then it is essential that the relationship between
the endpoint measured, the pharmacokinetic profile and
antitumour efficacy is understood. There is an opportunity
to do this in the pre-clinical setting, preferably using a
range of tumour models, but in addition early phase
clinical trials need to incorporate the ability to examine
this relationship into their design. Furthermore, it is
desirable to be able to compare the findings seen in the
clinic with the pre-clinical assessments to maximize the
understanding of this relationship, and thereby to come to
the appropriate decision following the early clinical trial
results.
From the above list of potential biological endpoints for
an antiangiogenic or antivascular drug, the desire to be
able to translate methodologies from pre-clinical experiments to early clinical trials drives some pragmatic choices
and raises issues about available imaging methods. The
range of imaging possibilities available for pre-clinical
experiments is generally wider than the range that is
readily available across a large number of potential clinical
sites.
Target inhibition
In the category of measurement of target inhibition
Bremer et al have elegantly demonstrated that MMP2 is
inhibited by 150 mg kg21 twice daily of prinomastat [13].
They used optical imaging in a mouse model with a probe
containing 2 fluorophores which in close proximity quench
the fluorescence. The probe is cleaved by MMP2, so the
fluorophores are no longer in close proximity and the
fluorescence emitted can be quantitated. Whilst this
technique is very helpful in the pre-clinical understanding
of the relationship of dose and plasma concentration to
antitumour effect it cannot be currently clinically applied.
This is due firstly to the problem of detecting emitted
fluorescence in deep seated tumours in patients, but also
due to the timeframe for the development and approval of
such an imaging probe. Imaging probes based on similar
principles for use with MRI are under development [14],
but for each probe specific to a particular protease a
separate approval would be necessary currently, and it is a
reflection of this problem that any such probes are only
likely to be available for clinical use after the vast majority
of MMPI are no longer in clinical development.
For VEGF pathway targeted therapies Collingridge et al
have used an 124I iodinated monoclonal antibody VG67e
which binds to human VEGF A for assessment of tumour
VEGF levels non-invasively [15]. Similarly HuMV833 has
been labelled with 124I allowing imaging of distribution
S83
S M Galbraith
into tumours [16]. However, there has not been any
comparison of tumour uptake by these methods with
measurement of VEGF level in tumour by alternative
methods so it is difficult to know whether the uptake of
the imaging probe into tumour tissue is solely affected by
the level of VEGF in tumour or if it is also affected by the
pharmacokinetic distribution of the probe itself.
Measuring changes in microvessel function
In the category of measuring changes in tumour
microvasculature there are a number of techniques
available in pre-clinical models that could be translated
into human studies. These include the use of low molecular
weight contrast medium enhanced dynamic contrastenhanced MRI (DCE-MRI), to measure transfer constant
(Ktrans), initial area under gadolinium curve (IAUC) and
leakage space (ve) [17]; high molecular weight contrast
agents to measure vessel permeability (KPS) and fractional
plasma volume (fPV); and BOLD MRI to measure T2*
relativity (R2*) which is sensitive to both blood oxygenation and blood flow [18, 19], and the change in BOLD
signal seen whilst breathing oxygen and CarbogenTM gas
to assess vessel maturity [20]. However, high molecular
weight contrast agents are not yet widely clinically
available, and the BOLD contrast method is dependent
on the field gradient used, making comparisons between
measurements made on different MR machines difficult.
PET imaging includes 15O labelled water for measurement
of blood flow, and 11C labelled carbon monoxide for
measurement of blood volume, although even with the
advent of microPET scanners, the use of 15O PET in preclinical models is unusual. Ultrasound techniques using
microbubble contrast agents have been developed for
measurement of blood flow, and have potential utility in
both pre-clinical and clinical settings [22, 23]. Therefore
from this list only a subset of techniques have been
available to be used both in pre-clinical studies and in the
clinical assessment of antiangiogenic or antivascular agents.
In a range of Phase I trials with endostatin, PET, CT,
ultrasound and DCE-MRI techniques were used, but there
was no clear pre-clinical data to help understand whether
the effect size seen (15O PET measured blood flow
averaged 20% reduction from baseline at low doses) was
relevant for antitumour efficacy or whether it was greater
than the change one might expect in the absence of
treatment and the cohort size per dose level was small [3,
4]. The data generated therefore are interesting and useful
for designing future studies, but without further studies
could not be used to drive decisions about dose or
schedule selection.
For HuMV833, a fully human anti-VEGF antibody,
Jayson et al have measured DCE-MRI changes across a
range of dose levels [6]. A median decrease of 44% in Kfp
(first pass permeability) was seen, but the authors comment was that when cohort sizes of three or less are used
in a trial with a range of different tumour types, that the
heterogeneity of response is such that discrimination of one
dose level from another becomes difficult. This is an
important point, and the advantage of using a single
tumour type in trials with an imaging endpoint is illustrated
in the data with Novartis’ (Basel, Switzerland) VEGFR
tyrosine kinase inhibitor, Vatalanib [5]. In a group of 22
patients with liver metastases from colorectal cancer who
S84
were treated across a dose range of 50 mg to 2000 mg daily,
there was a correlation seen between changes in DCEMRI parameter Ki after 2 days of dosing with change in
tumour size after 2 months. In addition a DCE-MRI
response to exposure relationship across the whole dose
range was determined, although again a relatively small
cohort size per dose level did not permit clear discrimination of one dose level from the next. Data relating the
effect size seen in patients to pre-clinical models have not
been published. Whilst the tolerability profile probably
also contributed to dose selection, the DCE-MRI data
helped to support the choice of 1200 mg daily, which is
being taken forward into Phase III trials in colorectal
cancer. The modelling of exposure–response using an
E-max model suggested a target 115 mM.h on day 2 was
needed to achieve a .40% decrease from baseline Ki. The
majority of patients who achieved this level of change
from baseline Ki had tumour regression at the end of cycle
2, although not sufficient regression to meet criteria for
partial response. At the 1200 mg daily dose the lower limit
of standard deviation was greater than the target exposure.
In the Phase I trials with combretastatin A4 phosphate,
both PET and DCE-MRI were used [9, 11]. The results
with both techniques were broadly comparable, although
the PET technique has the advantage of determining an
absolute value for tumour and normal tissue blood flow,
whilst DCE-MRI measures a composite of blood flow,
vessel permeability and surface area. Whilst the exact
physiological effects are less clearly understood from
measurement of change in such a composite endpoint
the DCE-MRI results nevertheless generated valuable
information which contributed to Phase II dose selection
in addition to the toxicity profile. This is because there had
been pre-clinical work done comparing the dose response
with DCE-MRI to that with a technique for measuring
absolute blood flow in the same animal model [24], and an
understanding of how the effect size seen related to
antitumour efficacy when the drug is used in combination
therapy regimens [25, 26]. The effect size measured (mean
37% decrease in Ktrans from baseline) was greater than the
95% limits of change determined in reproducibility studies
from a cohort of patients without any treatment with
similar characteristics to those in the trial [27]. In addition,
there was a significant correlation of DCE-MRI change
within 24 h of dosing with drug exposure [9], although the
cohort size used was too small to allow direct comparisons
of one dose level with another.
Effects on cellular health
The third category of biomarkers is less specific to the
field of anti-angiogenesis and vascular targeting agents but
includes FDG and fluorine-18 labelled thymidine (FLT)
PET imaging techniques which are both available in the
pre-clinic as well as the clinic. These type of imaging
studies have been used to assess the effects of SU11248
which inhibits VEGFR2 as well as a number of other
tyrosine kinases [7]. Although significant effects (determined as greater than 20% change from baseline) were
seen following treatment with this compound, it is difficult
to determine if that is due to the VEGFR inhibition or to
inhibition of other kinases. It is of note though that the
response rate on FDG-PET scanning in patients with
Gleevec (Novartis) resistant GIST tumours was .70%,
The British Journal of Radiology, Special Issue 2003
Angiogenic imaging biomarkers
whereas the response rate on CT scan criteria was only
10%. Studies with Gleevec in patients with GIST have
demonstrated a better correlation of survival with the FDG
changes post treatment, than for CT assessed response.
Incorporating imaging into clinical trials
The combined information from these clinical experiences using imaging in early clinical trials with such agents
can be used to develop some general conclusions. Rather
than having a small number of patients with varying levels
of change in the endpoint it is desirable to determine a
priori quantitative criteria for the minimum change in the
imaging endpoint needed to have confidence that a real
effect is being measured of relevance to the drug’s mechanism of action. To achieve this it is important both to
understand the relationship of change in imaging endpoint
to anti-tumour efficacy from pre-clinical or prior clinical
studies and to understand the reproducibility of the technique as used in the clinical trial. The reproducibility of
FDG-PET has been published by Weber et al. In their hands
the standard deviation of the difference between repeat
measurements was around 10% for both standardized uptake
value (SUV) and the rate constant for the uptake of FDG
into the tissue (K1). DCE-MRI which does not generally
determine the arterial input function has a within-patient
coefficient of variation (wCV) for Ktrans of 24% [10]. If a
power injector is used some improvement can be seen;
Lankester has reported wCV for Ktrans of 20% (pers.
comm.). The IAUC proposed by Evelhoch is more
reproducible with a wCV of 12% [27].
The magnitude of the change in imaging parameters at
dose levels which have antitumour efficacy either as a
single agent or in combination studies will vary with the
individual drug. For CA4P the minimum change in
tumour vascular volume associated with significant antitumour efficacy in pre-clinical models was 40% (Dr S Hill,
pers. comm.). For vatalanib, the clinical data from
Morgan et al [5] also suggest a minimum 40% decrease
from baseline was associated with tumour regression in
patients with liver metastases from colorectal cancer.
Using the reproducibility data above one can determine
that for Ktrans a cohort size of 14 patients would be needed
to measure a decrease of 15% with statistical significance,
for IAUC a cohort of 9 patients would be required for this
effect size. Clearly to have this size of cohort across
all cohorts of a dose escalation study is impractical. It
therefore seems more logical to obtain the safety and
pharmacokinetic information across cohorts of 3 as is
standard practice in Phase I and then to expand 2 or 3
cohorts at dose levels which are well tolerated to a size of
10–15 patients depending on the chosen imaging methodology and effect size of interest. Selecting a single tumour
type for this imaging phase will also be likely to improve
the consistency of the response seen.
There are other practical implications of the above
analysis. In order to accrue 30–45 patients for detailed
imaging studies within an acceptable time period, multiple
sites will be required. The imaging technology used has
therefore to be reasonably widely available with the sites
chosen having prior experience in its use. It is not always
possible to combine datasets from imaging equipment
made by different manufacturers so this needs to be
determined prior to site selection. It is clearly preferable
The British Journal of Radiology, Special Issue 2003
that imaging data from all sites should be combined in one
database, so sites need to be able to agree on a common
imaging protocol, and adhere to it. Quality assurance in as
near real time as possible is important.
Another issue that does not generally receive as much
attention as it needs is the determination of how regions of
interest (ROIs) are chosen. Even for higher resolution
methodologies such as CT and MRI this is a key aspect
that can significantly affect reproducibility of measurements. In different studies outlining of the whole tumour,
choosing a ROI ‘‘within the area of highest contrast agent
uptake’’ or exclusion of ‘‘necrotic’’ areas of tumour have
been used. Since choosing an ROI within a region of high
uptake is more likely to be affected by subjective decisions,
this should be avoided. Preferably one person should
be responsible for ROI selection throughout a study and
interobserver and intraobserver variability should be
known or measured.
These practical issues again have relevance to the choice
of imaging endpoint. Although absolute measures of vessel
permeability, blood flow and blood volume would be
highly desirable from the perspective of understanding the
physiological changes occurring within a tumour as a
result of treatment with an antiangiogenic or antivascular
drug, if the endpoint used is less clearly defined physiologically but is still correlated with antitumour efficacy, is
readily measurable at a range of clinical sites and has good
reproducibility then it may well be a better choice. For
DCE-MRI both Ktrans/Ki and IAUC proposed by
Evelhoch are good examples, and for PET the SUV of
FDG has proved to be useful in the trials with Gleevec
and SU11248. Nevertheless it is still important to understand the limitations on interpretation which must be
made when such end-points are chosen.
In conclusion, we now have available to us a range of
imaging techniques which are beginning to influence
decisions in early clinical development of antiangiogenic,
antivascular and signal transduction inhibiting agents. In
order to maximize their utility, clinical trial design needs to
be adapted to accommodate such endpoints rather than
using imaging as an ‘‘optional’’ add on to a standard design.
References
1. Zucker S, Cao J, Chen WT. Critical appraisal of the use of
matrix metalloproteinase inhibitors in cancer treatment.
Oncogene 2000;19:6642–50.
2. Hidalgo M, Eckhardt SG. Development of matrix metalloproteinase inhibitors in cancer therapy. J Natl Cancer Inst
2001;93:178–93.
3. Herbst RS, Mullani NA, Davis DW, Hess KR, McConkey
DJ, Charnsangavej C, et al. Development of biologic markers
of response and assessment of antiangiogenic activity in a
clinical trial of human recombinant endostatin. J Clin Oncol
2002;20:3804–14.
4. Thomas JP, Arzoomanian RZ, Alberti D, Marnocha R, Lee
F, Friedl A, et al. Phase I pharmacokinetic and pharmacodynamic study of recombinant human endostatin in patients
with advanced solid tumors. J Clin Oncol 2003;21:223–31.
5. Morgan B, Drevs J, Steward W, DiLea C, Lee L, Marme D,
et al. Dynamic contrast-enhanced magnetic resonance imaging as a biomarker for the pharmacological response of
PTK787/ZK222584, an inhibitor of the vascular endothelial
growth factor receptor tyrosine kinases, in patients with
advanced colorectal cancer and liver metastases: results from
two Phase I studies. J Clin Oncol 2003;21:3955–64.
S85
S M Galbraith
6. Jayson GC, Zweit J, Jackson A, Mulatero C, Julyan P,
Ranson M, et al. Molecular imaging and biological evaluation of HuMV833 anti-VEGF antibody: implications for trial
design of antiangiogenic antibodies. J Natl Cancer Inst 2002;
94:1484–93.
7. Toner G, Mitchell P, De Boer R, Gibbs P, Hicks R, Scott A,
et al. PET imaging study of SU11248 in patients with
advanced malignancies. Abstract Am Soc Clin Oncol.
Chicago, USA; 2003.
8. O’Farrell AM, DePrimo SE, Manning WC, Bello C, Raleigh
J, Hicks R, et al. Analysis of biomarkers of SU11248 action
in an exploratory study in patients with advaced malignancies. Abstract Am Soc Clin Oncol. Chicago, USA; 2003.
9. Galbraith S, Maxwell R, Lodge M, Tozer G, Wilson J,
Taylor N, et al. Combretastatin A4 phosphate has tumor
antivascular activity in rat and man as demonstrated by dynamic
magnetic resonance imaging. J Clin Oncol 2003;21:2831–42.
10. Galbraith SM, Rustin GJ, Lodge MA, Taylor NJ, Stirling JJ,
Jameson M, et al. Effects of 5,6-dimethylxanthenone-4-acetic
acid on human tumor microcirculation assessed by dynamic
contrast-enhanced magnetic resonance imaging. J Clin Oncol
2002;20:3826–40.
11. Anderson H, Yap J, Miller M, Robbins A, Jones T, Price P.
Assessment of pharmacodynamic vascular response in a
Phase I trial of combretastatin A4 phosphate. J Clin Oncol
2003;21:2823–30.
12. DelProposto Z, LoRusso P, Latif Z, Morton P, Wheeler C,
Barge A, et al. MRI evaluation of the effects of the vascular
targeting agent ZD6126 on tumor vasculature. Abstract Am
Soc Clin Oncol Orlando, Florida; 2002.
13. Bremer C, Tung CH, Weissleder R. In vivo molecular target
assessment of matrix metalloproteinase inhibition. Nature
Med 2001;7:743–8.
14. Josephson L, Kircher MF, Mahmood U, Tang Y, Weissleder
R. Near-infrared fluorescent nanoparticles as combined MR/
optical imaging probes. Bioconjug Chem 2002;13:554–60.
15. Collingridge DR, Carroll VA, Glaser M, Aboagye EO,
Osman S, Hutchinson OC, et al. The development of
[(124)I]iodinated-VG76e: a novel tracer for imaging vascular
endothelial growth factor in vivo using positron emission
tomography. Cancer Res 2002;62:5912–9.
16. Jackson A, Haroon H, Zhu XP, Li KL, Thacker NA, Jayson
G. Breath-hold perfusion and permeability mapping of hepatic malignancies using magnetic resonance imaging and a firstpass leakage profile model. NMR Biomed 2002;15:164–73.
S86
17. Tofts P, Brix G, Buckley D, Evelhoch J, Henderson E,
Knopp M, et al. Estimating kinetic parameters from dynamic
contrast-enhanced T(1)-weighted MRI of a diffusable tracer:
standardized quantities and symbols. J Magn Reson Imaging
1999;10:223–32.
18. Griffiths JR, Taylor NJ, Howe FA, Saunders M, Robinson
SP, Hoskin PJ, et al. The response of human tumors to
carbogen breathing, monitored by gradient-recalled echo
magnetic resonance imaging. Int J Radiat Oncol Biol Phys
1997;39:697–701.
19. Robinson SP, Howe FA, Griffiths JR. Noninvasive monitoring of carbogen-induced changes in tumor blood flow and
oxygenation by functional magnetic resonance imaging. Int
J Radiat Oncol Biol Phys 1995;33:855–9.
20. Neeman M, Provenzale JM, Dewhirst MW. Magnetic
resonance imaging applications in the evaluation of tumor
angiogenesis. Semin Radiat Oncol 2001;11:70–82.
21. Neeman M, Dafni H, Bukhari O, Braun RD, Dewhirst MW.
In vivo BOLD contrast MRI mapping of subcutaneous
vascular function and maturation: validation by intravital
microscopy. Magn Reson Med 2001;45:887–98.
22. Leong-Poi H, Christiansen J, Klibanov AL, Kaul S, Lindner
JR. Noninvasive assessment of angiogenesis by ultrasound
and microbubbles targeted to alpha(v)-integrins. Circulation
2003;107:455–60.
23. Kim SW, Park SS, Ahn SJ, Chung KW, Moon WK, Im JG,
et al. Identification of angiogenesis in primary breast
carcinoma according to the image analysis. Breast Cancer
Res Treat 2002;74:121–9.
24. Maxwell RJ, Wilson J, Prise VE, Vojnovic B, Rustin GJ,
Lodge MA, et al. Evaluation of the antivascular effects of
combretastatin in rodent tumours by dynamic contrast
enhanced MRI. NMR Biomed 2002;15:89–98.
25. Tozer GM, Prise VE, Wilson J, Locke RJ, Vojnovic B,
Stratford MR, et al. Combretastatin A-4 phosphate as a
tumor vascular-targeting agent: early effects in tumors and
normal tissues. Cancer Res 1999;59:1626–34.
26. Chaplin DJ, Pettit GR, Hill SA. Anti-vascular approaches to
solid tumour therapy: evaluation of combretastatin A4
phosphate. Anticancer Res 1999;19:189–95.
27. Galbraith SM, Lodge MA, Taylor NJ, Rustin GJ, Bentzen S,
Stirling JJ, et al. Reproducibility of dynamic contrastenhanced MRI in human muscle and tumours: comparison
of quantitative and semi-quantitative analysis. NMR Biomed
2002;15:132–42.
The British Journal of Radiology, Special Issue 2003