Limits of [18F]-FLT PET as a biomarker of proliferation in oncology

Abstract

Background: Non-invasive imaging biomarkers of cellular proliferation hold great promise for quantifying response to personalized medicine in oncology. An emerging approach to assess tumor proliferation utilizes the positron emission tomography (PET) tracer 3'-deoxy-3'[(18)F]-fluorothymidine, [(18)F]-FLT. Though several studies have associated serial changes in [(18)F]-FLT-PET with elements of therapeutic response, the degree to which [(18)F]-FLT-PET quantitatively reflects proliferative index has been continuously debated for more that a decade. The goal of this study was to elucidate quantitative relationships between [(18)F]-FLT-PET and cellular metrics of proliferation in treatment naïve human cell line xenografts commonly employed in cancer research.

Methods and findings: [(18)F]-FLT-PET was conducted in human cancer xenograft-bearing mice. Quantitative relationships between PET, thymidine kinase 1 (TK1) protein levels and immunostaining for proliferation markers (Ki67, TK1, PCNA) were evaluated using imaging-matched tumor specimens. Overall, we determined that [(18)F]-FLT-PET reflects TK1 protein levels, yet the cell cycle specificity of TK1 expression and the extent to which tumors utilize thymidine salvage for DNA synthesis decouple [(18)F]-FLT-PET data from standard estimates of proliferative index.

Conclusions: Our findings illustrate that [(18)F]-FLT-PET reflects tumor proliferation as a function of thymidine salvage pathway utilization. Unlike more general proliferation markers, such as Ki67, [(18)F]-FLT PET reflects proliferative indices to variable and potentially unreliable extents. [(18)F]-FLT-PET cannot discriminate moderately proliferative, thymidine salvage-driven tumors from those of high proliferative index that rely primarily upon de novo thymidine synthesis. Accordingly, the magnitude of [(18)F]-FLT uptake should not be considered a surrogate of proliferative index. These data rationalize the diversity of [(18)F]-FLT-PET correlative results previously reported and suggest future best-practices when [(18)F]-FLT-PET is employed in oncology.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Figure 1. Thymidine salvage and de novo synthesis pathways.

In thymidine salvage, thymidine is transported across the cell membrane and phosphorylated by TK1 into thymidine monophosphate (TMP). The thymidine is further phosphorylated into thymidine diphosphate (TDP) and thymidine triphosphate (TTP) and then incorporated into DNA. Alternatively, using the de novo synthesis pathway, deoxyuridine monophosphate (dUMP) is converted to TMP by TS which can then be further phosphorylated and incorporated into DNA. Similarly to thymidine, [18F]-FLT is transported into the cell and phosphorylated into [18F]-FLT monophosphate ([18F]-FLTMP) and trapped by TK1. [18F]-FLTMP can be further phosphorylated into [18F]-FLT diphosphate ([18F]-FLTDP) and [18F]-FLT triphosphate ([18F]-FLTTP), however, due to the substitution of OH with 18F in the 5-prime position, [18F]-FLTTP is not incorporated into the DNA.

Figure 2. IHC markers of proliferation in HCT-116 and DiFi human CRC xenografts.

Representative high power microscopic images (40x) of HCT-116 and DiFi xenograft tissue stained for Ki67, PCNA, and TK1 are shown along with quantification of the percentage of positive cells per field for each marker. HCT-116 and DiFi xenograft tissues exhibit a similar proportion of proliferating cells as measured by Ki67 indices (68.50 ± 5.52 % vs. 72.37 ± 12.64%; p  =  0.0651). The proportion of PCNA positive cells, representing S-phase, was significantly increased in HCT-116 xenografts (53.91 ± 3.18 %) compared to DiFi xenografts (13.59 ±1.48%; p  =  0.0007). Similar to PCNA, the percentage of TK1-positive cells was significantly higher in HCT-116 xenografts (38.32 ± 1.90 %) than DiFi cells (17.95 ± 3.95 %; p  =  0.0007). TK1 indices for both cell lines were reduced compared to Ki67 indices.

Figure 3. HCT-116 xenografts exhibit higher [18F]-FLT flux than DiFi xenografts.

Representative time activity curves for HCT-116 (A) and DiFi (D) xenografts and left ventricle estimates of blood pool activity are shown. In both models, a sharp rise in activity in the blood pool is observed following tracer injection, followed by rapid clearance. Fit lines derived from compartmental modeling closely matched the measured data. Parametric maps reveal heterogeneous intratumoral delivery of [18F]-FLT in both HCT-116 (B) and DiFi (E) xenografts (tumor localized by arrowhead). Increased [18F]-FLT delivery was observed in the urinary bladder (denoted by UB) in both cell line xenografts. The tumor regions exhibiting the greatest [18F]-FLT flux in HCT-116 tumors (C) also exhibited the highest tracer delivery. Only modest [18F]-FLT flux was observed in DiFi xenografts (F), despite having similar delivery as HCT-116 xenografts (B).

Figure 4. Static [18F]-FLT PET uptake is 2-fold greater in HCT-116 xenografts than DiFi xenografts.

Representative 20 minute static PET scans reflect the relative 60 minute accumulation of [18F]-FLT (%ID/g) qualitatively and show increased tracer uptake in HCT-116 xenografts (A) compared to DiFi (B) xenografts (tumor localized by arrowhead). Spatially, %ID/g maps (A,B) are similar to KFLT maps derived from parametric modeling (Fig. 3C,F). Static PET imaging reveals approximately a 2-fold increase in [18F]-FLT accumulation in HCT-116 xenografts (8.56 ± 1.17 %ID/g) compared to DiFi xenografts (3.92 ± 1.08 %ID/g; p < 0.0001) (C) and was similar to [18F]-FLT flux (KFLT) means derived from compartmental modeling (Table 1).

Figure 5. Isogenically matched cell line xenografts illustrate that de novo pathway utilization results in decreased [18F]-FLT uptake.

Western blot analysis illustrated reduced TK1 levels and elevated TS levels in HCT-116p21-/- xenografts compared to wild type HCT-116 xenografts (A). Densitometry of the western blot illustrated that HCT-116p21-/- tumors exhibited one-third less TK1 protein and double the TS protein compared to the parental line (B). Illustrating the sensitivity of [18F]-FLT PET to de novo pathway utilization, PET imaging of HCT-116 (C) and HCT-116p21-/- (D) xenografts closely reflected the relative tumor cell TK1 levels inherent to each model (tumor localized by arrowhead). HCT-116 xenografts (8.56 ± 1.17 %ID/g) exhibited approximately 1/3 greater uptake than analogous HCT-116p21-/- xenografts (6.91 ± 1.07 %ID/g; p  =  0.005) (E).