AZD1152 is effective in inhibiting the growth of two human colorectal/colonic cancer cell lines (HCT116 and SW620) in culture and xenografts in nude mice (9
) ( and ). Both cell lines were sensitive to AZD1152-HQPA in the low nano-molar range. A robust treatment response based on tumor volume measurements was clearly obtained in HCT116 xenografts using a weekly schedule of i.p. AZD1152 administration (100 mg/kg i.p. injections daily on two consecutive days per week), a dose schedule previously reported to be effective in mice (9
). The treatment response of SW620 xenografts was somewhat delayed and less robust compared to the HCT116 xenograft response.
One question we asked was whether [18
F]FDG or [18
F]FLT PET imaging could predict treatment response “early”, within several days or weeks of initiating treatment. Since hematologic toxicity is observed with prolonged administration of AZD1152 (31
), early assessment of treatment response could be important in at least two respects. First, if it is possible to identify a non-responding tumor “early” by [18
F]FDG or [18
F]FLT PET imaging, would the imaging results be sufficient criteria for discontinuing AZD1152 treatment before hematologic toxicity develops? Second, could PET imaging be used to establish the minimum effective dose of AZD1152 in individual patients and thereby reduce hematologic toxicity during prolonged administration of the drug?
F]FDG PET imaging results were surprising given the tumor volume treatment responses that were observed in both xenograft models. Sequential FDG PET imaging results showed little or no difference in FDG uptake between AZD1152-treated and non-treated xenografts, when the radioactivity data are expressed as maximum-voxel values. [FOOTNOTE: Measures based on maximum-voxel values are used to avoid partial volume imaging effects for small tumors; measures based on region of interest (ROI) measurements are usually used for larger tumors
.] These results suggested that glucose utilization by HCT116 and SW620 tumor cells is not significantly influenced by AZD1152 treatment, although treatment does have a significant effect on tumor growth and volume. However, this finding was inconsistent with our understanding of the mechanism of AZD1152 anti-tumor effects (9
). The major regulatory pathways of glucose metabolism are largely mediated through insulin receptor (IR) signaling and the AMP activated protein kinase signaling pathways (24
). Since the IR and the AMP signaling pathways are activated by Aurora Kinases, selective inhibition of Aurora Kinases (and Aurora Kinase B, in particular) was expected to reduce tumor glucose utilization.
When the FDG results are expressed as total tumor metabolism (or radioactivity) as has previously been suggested (15
), a highly significant AZD1152 treatment effect was observed in both HCT116 and SW620 xenografts. However, the total tumor metabolic response observed in this study largely reflects differences in tumor volume between treated and non-treated tumors, not a change in glucose metabolism of tumor cells within an imaging voxel. The absence of a drug-effect on maximum-voxel FDG uptake is more clearly appreciated when the xenograft values were normalized to that of the surrounding, non-tumor tissue; this normalization accounts in part for inter-animal variations and FDG input function differences. Thus, FDG PET may not be a useful paradigm for non-invasive monitoring of AZD1152 treatment-response, at least as reflected in these two animal xenograft models.
The [18F]FLT PET imaging studies also yielded surprising results. SW620 xenografts showed little or no accumulation of [18F]FLT above background levels and there was no difference in [18F]FLT accumulation between AZD1152-treated and non-treated SW620 xenografts. These results conflicted with the robust Ki67 staining pattern observed in untreated SW620 xenografts. In contrast, [18F]FLT uptake in untreated HCT116 xenografts was 10-fold higher than that in untreated SW620 xenografts, and this difference was 37-fold when the values are corrected for background radioactivity. The [18F]FLT imaging-response pattern following AZD1152 treatment of HCT116 xenografts was also quite different. [18F]FLT uptake decreased to 20% or less of that measured in non-treated HCT116 xenografts over the three week treatment period, and these imaging results were consistent with the decrease in Ki67 staining following AZD1152 treatment.
F]FLT, as well as [11
C]TdR) tumor proliferation imaging depends on two major components: i) transport of pyrimidine nucleotides across cell membranes and ii) the activity of thymidine kinase. Both components are highly regulated and the expression levels of both transporter and kinase depend on the cell cycle and rate of cell proliferation (28
). Another determinant, which is sometimes ignored in [18
F]FLT and [11
C]TdR PET imaging studies, is an assessment of whether the endogenous pathway of thymidine synthesis is active in the cell or whether exogenous thymidine is the dominant source of the nucleotide for DNA synthesis via the “salvage pathway” (31
To further test whether the de novo
pathway of thymidine synthesis was dominant in SW620 cells, but not in HCT116 cells, we performed four different in vitro
, we found an 8-fold difference in the rates [14
C] thymidine and [3
H]FLT accumulation in HCT116 and SW620 cells in culture, , which demonstrated an impairment of tracer uptake (the combination of transport and phosphorylation) of these compounds in SW620 cells. This suggests that utilization of exogenous thymidine as a salvage mechanism may be less predominant in the SW620 cell line. Second
, we found only a small (1.5-fold) difference in the exponential doubling time between the two cell lines. Third
, we tested the sensitivity of HCT116 and SW620 cells to MTX and 5FU and found a marked difference between the two cell lines consistent with prior studies (32
, we performed immunoblots for thymidine kinase and thymidylate synthetase. The immunoblots for TK and TS protein showed slightly lower levels of TK in SW620 cells, but this difference was not considered significant and may not be an accurate measure of intracellular enzymatic activity. In total, these in vitro
results are consistent with the imaging results described above, and are also consistent with SW620 cells relying mainly on the de novo pathway of TdR synthesis. Although HCT116 cells do incorporate TdR at a higher rate than SW620 cells, consistent with the tumor imaging studies, it is not clear if the lower EC50 values for MTX and 5FU in the SW620 cell line are attributable only to lower salvage of TdR in this cell line. Further studies will be required to more fully address this question. However, inhibition of Aurora Kinases A and B has been shown to result in the down-regulation of thymidine kinase 1 (TK1) in HCT116 cells via the Rb pathway (33
), through phosphorylation of histone H3 and by p53 protein stabilization and induction of p21, and this is consistent with our findings
The wide disparity in [18F]FLT PET imaging between untreated HCT116 and SW620 xenografts (37-fold, background-corrected) compared to a 2.3-fold difference in tumor doubling time, respectively, provides a striking example of potential limitations associated with [18F]FLT (and [11C]TdR) tumor proliferation imaging studies. We suggest that these results highlight the importance of determining the contribution of the de novo pathway of thymidine synthesis when [18F]FLT or [11C]TdR PET is used to image and monitor tumor proliferation. The PET studies image/measure the exogenous component of thymidine incorporation into DNA via the salvage pathway. When the fraction of thymidine synthesized via the de novo pathway and incorporated into DNA increases, the magnitude of [18F]FLT and [11C]TdR uptake in the tumor is correspondingly reduced.
Although this issue is frequently ignored in clinical studies using [18
F]FLT or [11
C]TdR PET to image and measure tumor proliferation, the initial studies establishing the tumor-proliferation imaging paradigms clearly identified de novo
thymidine synthesis as a potential limitation and a confounding factor for quantitation of proliferation (31
). A review of multiple publications (35
) and more than 300 cancer patients imaged with [18
F]FLT demonstrated that [18
F]FLT was accumulated above background in the vast majority of the reported imaging studies. These studies included a preponderance of lung/thoracic tumors (130 patients), followed by glioma (n=52), lymphoma, pancreatic, breast and colon tumors (n=17–34), as well as a smaller numbers of other cancers. Nevertheless, “false negative” [18
F]FLT imaging results are recorded in about 10% of patient studies, and we suspect that this percentage is actually higher than what has been reported in the literature. For example, 2/23 patients with non-small cell lung cancer, 3/9 patients with lung metastases (2 colorectal, 1 melanoma), and 1/1 pulmonary carcinoid who had [18
F]FLT imaging were classified as “false negative” scans (51
). Other studies have noted that some tumors imaged well with [18
F]FDG, but not with [18
F]FLT or that some tumors with low FLT uptake had moderate Ki67 staining, despite an overall correlation between these two independent measures.
Finally, what remains unexplored in most [18
F]FLT drug-response monitoring studies is whether a particular drug therapy results in a shift in TdR utilization from the salvage pathway to the de novo
synthesis of thymidine or vise versa. This was recently demonstrated in a small number of patients (n=6) with breast cancer (52
). This cautionary note needs wider appreciation and requires further study.
Early-assessment of treatment response using [18F]FDG and [18F]FLT PET imaging has been reported in many clinical as well as preclinical studies. In this study we tested whether [18F]FDG and [18F]FLT PET imaging, early during the course of treatment with an Aurora Kinase B inhibitor - AZD1152, could be used to monitor treatment-response in two different colon cancer xenografts models. Both imaging probes demonstrated a response to AZD1152 therapy, but in quite different ways. These differences are likely to be important considerations in the assessment treatment response using [18F]FDG and [18F]FLT PET imaging in clinical studies as well. For example, maximum-voxel values are a commonly used measure of tumor metabolic activity to reduce partial volume errors, whereas total tumor radioactivity/metabolism also takes into account changes in tumor volume. There was a significant difference between these two measures with [18F]FDG in both HCT116 and SW620 xenografts. Although [18F]FLT imaging showed a robust response in AZD1152-treated HCT116 xenografts, this was not seen in AZD1152-treated SW620 xenografts, despite prominent Ki67 staining of the tumor cells. This study presents additional data that suggests thymidine incorporation into the DNA of SW620 cells occurs predominantly through the de novo pathway of thymidine synthesis, and highlights the importance of recognizing the relative contributions of both the de novo and salvage pathways of thymidine utilization when [18F]FLT (or [11C]thymidine) PET imaging is used to image tumor proliferation or response to therapy.