Technical validation of quantitative methods for 3 'deoxy-3-[18F] fluorothymidine ([18F] FLT) PET in liver metastases of colorectal cancer patients

Principle of positron emission tomography (PET)
PET is a non-invasive imaging technique based on the use of biologically
relevant compounds labelled with short-lived positron-emitting radionuclides
such as carbon-11, nitrogen-13, oxygen-15 and fluorine-18. In clinical
applications, a very small amount of labelled compound (radiopharmaceutical or
radiotracer) is introduced into the patient usually by intravenous injection
and the concentration of tracer in tissue is measured using the scanner. During
its decay process, the radionuclide emits a positron which, after travelling a
short distance (1-2 mm), encounters an electron from the surrounding
environment. The two particles combine and "annihilate" each other resulting in
the emission in opposite directions of two γ-rays of 511 keV each. The image
acquisition is based on the external detection in coincidence of the emitted γ-
rays, and a valid annihilation event requires a coincidence between two
detectors on opposite sides of the scanner. For accepted coincidences, lines of
response connecting the coincidence detectors are drawn through the object and
used in the image reconstruction.

[18F]fluorothymidine (FLT): measuring tumor cell proliferation
The most widely used PET tracer, and the only one currently approved by the US
Food and Drug Administration, is [18F] FDG (2-fluoro-2-deoxy-D-glucose), which
canbe used in the diagnosis, staging, treatment monitoring and radiotherapy
planning of a number of cancers. However, since FDG is not tumor specific and
both false-positive and false- negative results are common, there is ongoing
research into possible alternatives, including [18F] FLT [1]. [18F] FLT-PET is
widely investigated as a proliferation marker in oncology. [18F] FLT follows
the salvage pathway of endogenous thymidine in the cell. However, unlike
thymidine, [18F] FLT is trapped in the cytosol and is not incorporated into
DNA. This process supplements the pool of thymidine monophosphate provided by
de novo synthesis. After intravenous injection, [18F] FLT enters tumor cells
both via a nucleoside transporter and partly via passive diffusion. Inside
proliferating cells, FLT is accepted as substrate by thymidine kinase 1
(TK-1), which phosphorylates it, thereby trapping it in cells.
FLT-monophosphate is further phosphorylated to di- and triphosphate forms.
Phosphorylation by TK-1 is the rate limiting step in FLT accumulation in
proliferating cells, causing FLT to accumulate in proportion to TK-1 activity
[2]. Published data on [18F] FLT-PET are heterogeneous and it is not clear to
what extent this relates to different pharmacokinetic characteristics,
biological changes, image resolution, or quantification methods. To date, the
majority of clinical research studies of FLT-PET in humans have focused on
validating it as an accurate measurement of proliferation in a broad spectrum
of cancers,
and assessing its usefulness for diagnosing and staging malignancies. However,
FLT undergoes predominantly hepatic metabolism to form FLT-glucuronide and
therefore shows high physiological uptake in the liver, with the result that
tumors (metastases) in the liver can prove difficult to identify. The accurate
detection of proliferation of liver tumors with FLT is therefore a clinically
relevant problem. There are 2 approaches to image analysis with [18F] FLT PET.
One simple method, which is practical in the clinic uses the standardized
uptake value
(SUV) derived from static images. The second method uses dynamic image
sequences and kinetic modeling of tracer uptake. Pharmacokinetic (pk) modelling
of PET tracers is the gold standard but it requires arterial blood sampling and
dynamic scanning. In case of [18F] FLT, the net uptake rate constant of [18F]
FLT, Ki, determined by non-linear regression (NLR) of an irreversible
two-tissue compartment model is typically used as the gold standard. Two
simplified methods often are used to (semi-)quantitatively assess [18F] FLT
uptake: graphical (Patlak) analysis [3] and standardized uptake values (SUV).
Patlak analysis assumes irreversible trapping in tissue, and its accuracy thus
depends on the assumption that no significant dephosphorylation occurs within
the time course of the study. Both NLR and Patlak measure net uptake of [18F]
FLT, taking into account the concentration of tracer in plasma during the
course of the study. Only NLR, however, allows for measurements of individual
rate constants between compartments and for an implicit correction for blood
volume in the tissue of interest. SUV is the ratio of tissue concentration and
injected activity at a certain time after administration of the tracer. It does
not take tracer kinetics into account, but has the advantage that it is a
single scan procedure that does not require plasma data. In daily clinical
practice, a static PET scan in whole body mode is the preferred clinical
procedure. In this context only SUV is feasible. Alternatively, kinetic
filtering methods have been proposed for the detection of liver tumors,
requiring procedures of similar complexity [4]. This procedure is not suited
for daily clinical practice and it is not suited for whole body acquisitions.

Detection of liver metastases with FLT-PET
Francis and coworkers reported imaging 17 patients with primary or CRCs, all 6
peritoneal metastases and 5 of 6 lung metastases. However, only 11 of 32 liver
metastases were detected because of high background activity in the liver,
caused by glucuronidation of FLT [5]. Zhou et al. investigated [18F] FLT PET/CT
in pretreatment evaluation of metastatic gastric cancer and compared it to
[18F] FDG PET/CT. In the liver and the bone marrow, sensitivity of FLT PET/CT
versus FDG PET/CT for detecting liver metastases and bone metastases was 30%
(6/20) versus 100% (20/20) and 1/5 (20%) versus 5/5 (100%), respectively (P<
0.05). This group concluded that FLT PET/CT imaging is not recommended for
pretreatment assessment of metastatic gastric cancer as it is not competent
enough to evaluate liver and bone metastases, moreover the high background
hepatic uptake may also cover gastric primary tumors located adjacent to the
liver. However, SUVmax were calculated for both radiotracers and no other
filtering methods were applied [6]. Gray et al. enhanced visualization of
tumors imaged with FLT-PET by using nonlinear kinetic filtering. A
classification algorithm was developed to isolate cancerous tissue from healthy
organs and was validated using 29
scans from patients with locally advanced or metastatic breast cancer. First,
metastatic CRC using FLT-PET. FLT-PET successfully detected all 6 primary
dose-normalized average time versus radioactivity curves for the major tissue
types were generated, and each image voxel was then classified according to the
tissue type it most likely to represent based on a comparison of its time
profile with those of the predefined classes. Images of only voxels likely to
show tumor tissue were then produced, by setting the intensities of all other
voxels to zero. Success of the filter in removing signal from healthy tissue,
whilst retaining that from tumor tissue, was assessed both qualitatively and
quantitatively. Reliability was quantified using test-retest data. A large
reduction in signal from the liver of 80% was observed following application of
the kinetic filter, whilst the majority of signal from both primary tumors and
metastases were retained. A scan acquisition of 60 minutes showed to be
sufficient to obtain the necessary results [2].

Post therapy FLT-PET
For simplified uptake measures to be valid for monitoring response or
predicting outcome, their relationship with the more accurate outcome
measures of full kinetic analysis must be similar before and after therapy [7].
However, systemic therapy might alter the correlations between NLR, Patlak and
SUV, as has previously

To investigate lesion detection in patients with liver metastases of colorectal
cancer with [18F] FLT-PET compared to CT at baseline (= gold standard). This
investigation will be done based on a qualitative assessment of the scans
([18F] FLT-PET and CT).

Secondary objectives:
1. To correlate the kinetic model measurements and simplified measurements from
the [18F] FLT PET
2. To determine test-retest reproducibility of quantitative assessment of liver
metastases* [18F] FLT uptake
3. To achieve a clinically feasible protocol f.e for dynamic PET, scan
durations < 30 minutes
4. To investigate change in FLT tumor uptake after two cycles of therapy and
compare it against a radiological response (RECIST) as gold standard measured
after 6 cycles of therapy

Prospective observational monocenter, multinational study on patients with mCRC
(liver metastases > 20 mm in diameter) will be scanned with [18F] FLT-PET on
2-3 separate occasions. Double baseline assessments (testretest) of FLT-PET
should be done within 1 week prior to first drug administration, and the
interval of test-retest should be between 24 hours (decay of [18F]) to 7 days.
During the therapy period, an optional third FLT PET scan could be performed at
the end of the second cycle (f.e. after 4 weeks for Folfox therapy). A
diagnostic CT will be performed together with the first and the third FLT PET
scan. The dynamic PET scans centered over the liver will be performed on a
PET-CT scanner. During PET, venous samples will be taken at different time
points. Dedicated in-house developed software will be used to quantify
kinetics. Personal and tumor characteristics will be registered (age, sex, body
weight and height, co-medication). Patients will be recruited by the Antwerp
University Hospital, Belgium (Prof. Dr. Stroobants S, Prof. Dr. Peeters M),
VUmc, Amsterdam, The Netherlands (Prof. dr. Hoekstra O.S.), UMC St Radboud
Nijmegen, The Netherlands (Prof. Dr. Oyen W), Manchester Cancer Research
Centre, Manchester, UK (Prof. Dr. Jackson A).

A PET scan is a regular diagnostic imaging technique. Each study will be
conducted in compliance with the radiation safety guidelines of the
department. Based on results we obtained from biodistribution studies in rats,
whole body radiation after intravenous injection of 300 MBq [18F] FLT is
approximately 6.5 mSv, including the low dose CT used for attenuation
correction. Since patients will undergo upto 3 FLT-PET scans, this will result
in a radiation dose of 19.5 mSv for the experimental component of the study. In
addition, the patient will undergo diagnostic CT scans as part of the standard
of care (8-9 mSv per scan) To compare, every person living in Belgium receives
a natural background radiation dose of 2.5 - 3 mSv per year. The maximum annual
amount of radiation allowed for a radiation worker such as a CT technologist or
radiologist, equals 20 mSv per year. The results of this study may have great
clinical benefit in using [18F] FLT PET-CT as drug monitoring tool in the
future, improving personalized therapy strategies for cancer patients. We
therefore consider the additional radiation burden acceptable.