In this study, we explored whether PET imaging of tumor proliferative activity using [18F]FLT can provide specific treatment response assessments in patients with soft tissue sarcoma. Furthermore, we investigated whether imaging findings reflected tumor proliferative activity as determined by immunohistochemical assays.
We found that tumor [18
F]FLT uptake after treatment was poorly but significantly correlated with the extent of posttreatment tumor necrosis, but was unrelated to TK1 expression and tumor proliferative activity by Ki-67 staining. The [18
F]FLT uptake decreased by more than 20% in 17 of 20 patients () and by more than 60% in 10 of 20 patients after completion of neoadjuvant therapy. However, only 3 of these 10 patients were classified as histopathologic responders. Changes in tumor uptake of [18
F]FLT were poorly but significantly correlated with histopathologic response ( and ). Although there is no reason to expect, biologically, that a measure of cellular proliferation would correlate with cell death, it is reasonable to test this correlation from a clinical standpoint, because earlier studies have shown that the degree of tissue necrosis after completion of neoadjuvant sarcoma therapy has prognostic value for long-term outcome.37
Furthermore, FLT uptake was unrelated to TK1 and Ki-67 expression in excised tissue (). These observations suggest that measurements of [18F]FLT tumor uptake after neoadjuvant chemotherapy and/or radiation therapy are unlikely to improve treatment response assessment as compared with [18F]FDG-PET.
The current study population was heterogeneous with regards to tumor histology and treatment approaches. This problem is not unique to sarcomas, because many other cancers also have many subtypes with varying biological characteristics and differing treatment approaches. However, to account for treatment heterogeneity, we conducted a subgroup analysis that excluded patients with GIST, those undergoing radiotherapy only, as well as those who received an angiogenesis inhibitor. No correlations between posttreatment tumor [18F]FLT uptake and Ki-67 or TK1 expression were found in this subgroup. Therefore, treatment heterogeneity did not appear to have an effect on the correlation between post-treatment imaging findings and immunohistochemistry.
To address the issue of disease heterogeneity, we conducted another analysis that only included the largest sarcoma subgroup (sarcomas NOS, n = 5). Here, TK1 expression but not Ki-67 expression was significantly correlated with posttreatment tumor [18F]FLT uptake. Thus, posttreatment [18F]FLT uptake might follow the expected correlation with TK1 in some sarcoma subtypes.
Imaging tests that are useful for response assessment only in subgroups of soft tissue sarcomas (or any other cancers) are, in our view, of limited clinical value. Any imaging biomarker proposed or used for therapy response assessments must be applicable across a wide spectrum of cancers, and within cancers, across a variety of genomic alterations and therapeutic approaches. This wide applicability has made [18F]FDG imaging successful for therapeutic response assessments.
We show in the current study that most sarcomas have a strong [18F]FLT phenotype at baseline (ie, are “routinely [18F]FLT avid”). If [18F]FLT should prove useful for 1) providing readouts of proliferative activity and 2) changes in proliferative activity in response to treatment, the underlying disease subtype as well as the treatment strategy should be irrelevant.
F]FLT was introduced as an indirect marker of tumor cell proliferative activity, the target enzyme being the S-phase–dependent TK1. In other words, [18
F]FLT imaging is supposed to image the proliferative tumor phenotype regardless of underlying tumor category. Because tumor [18
F]FLT uptake correlates with pretreatment Ki-67 expression, its use for monitoring the therapeutic effects on proliferative activity has been promoted. In the current study, we used an immunohistochemical approach to stain for TK1 protein. This approach is justified because TK1 protein levels are correlated with tumor proliferative activity38
and TK1 enzyme activity.39
However, even though TK1 activity is considered to be the key determinant of [18
F]FLT uptake, other factors such as up-or down-regulation of the equilibrative nucleoside transporter 1 have been shown to significantly alter [18
F]FLT uptake in vivo.40,41
Such altered equilibrative nucleoside transporter 1 expression and/or activity might have contributed to the lack of correlation between tumor FLT uptake and TK1 or Ki-67 expression.
Because ineffective treatment approaches can be modified, early therapy response assessments are important for managing patients with cancer. Radiation and chemotherapy, if successful, decrease tumor cell proliferation rates and/or kill tumor cells rapidly.42
These changes precede reductions in tumor size as determined by the Response Evaluation Criteria In Solid Tumors (RECIST) criteria,43
and it is clinically important to detect such changes as early as possible after the start of treatment.
Most clinical studies reported a significant correlation between Ki-67 expression and [18
F]FLT tumor uptake across many types of cancers in patients who were not recently treated.9–13
The relationship between TK1 protein levels or enzymatic activity and [18
F]FLT uptake has not been investigated extensively. However, in a study of 17 patients with oral cancer, pretreatment [18
F]FLT tumor uptake did not correlate with tumor TK1 antibody staining.24
Nevertheless, the lack of correlation between [18
F]FLT uptake in sarcoma and TK1 and Ki-67 expression in the current study was surprising. Animal experimental studies have shown that [18
F]FLT tumor uptake decreases rapidly in response to a variety of therapeutic interventions,15–17,44
and some have reported earlier response predictions with [18
F]FLT than with [18
Most clinical studies have shown marked decreases in tumor [18
F]FLT uptake in response to a variety of treatments. However, the predictive value of these changes was inconsistent. In patients with glioblastoma, the degree of reductions in [18
F]FLT uptake at 1 to 2 weeks after start of treatment was predictive of long-term survival.19
In breast cancer, reductions in [18
F]FLT uptake after 1 cycle of chemotherapy correlated with late changes in tumor markers and tumor size.20
Reductions in [18
F]FLT uptake 1 week after start of treatment were predictive of progression-free survival in patients with lung cancer who were treated with gefitinib.21
F]FLT responses with increased tumor uptake (flare) in 1 of 5 patients was observed in response to radiation treatment in patients with lung cancer.22
Less encouraging findings were reported in patients with rectal cancer who were imaged with [18
F]FLT 2 weeks after the start of neoadjuvant chemoradiation treatment.23
In this study, histopathological responders and nonresponders, who are defined by the fraction of necrotic tissue in excised tumor tissue, showed comparable decreases in [18
F]FLT tumor uptake. A uniform and therefore possibly nonspecific decrease in [18
F]FLT uptake was also observed in response to radiation therapy in patients with head and neck cancer24
and in patients with lymphoma early after start of chemotherapy.25
In another study of patients with lymphoma, posttreatment [18
F]FLT tumor uptake did not provide more accurate prognostic information than [18
Finally, neither changes in [18
F]FLT nor [18
F]FDG uptake in response to treatment predicted histopathological responses in patients with metastatic germ cell tumors.27
Several factors might account for the variable predictive value of [18
F]FLT imaging. First, the magnitude of changes in [18
F]FLT uptake might be dependent on treatment and/or tumor factors, as discussed above. In the current study, 13 of the 20 patients received ifosfamide-or gemcitabine-based chemotherapies.45
Both drugs exert complex effects on cellular deoxyribonucleotide pools, which, by feedback mechanisms, can affect TK1 enzymatic activity and/or transporter mechanisms that affect tumor [18
Second, 11 patients (55%) received radiation treatment that is known to reduce tumor vascularization and perfusion,46
which could also reduce tumor [18
F]FLT uptake. Another potential explanation considers leakage of TK1 and/or thymidine in the tumor microenvironment, which may also interfere with tumor [18
In addition, efflux pumps associated with chemotherapy resistance might also affect tumor [18
F]FLT uptake after treatment.49
F]FLT-MP dephosphorylation by cytoplasmic nucleotidases might reduce [18
F]FLT retention by the tumor.
Finally, the lack of correlation might be unrelated to altered [18F]FLT kinetics in response to treatment. Ki-67 may not be an ideal marker of tumor proliferation after therapy. This protein is expressed throughout the cell cycle except during the G0 phase. Therefore, cell cycle arrest in any other phase in response to treatment should leave tumor cells positive for Ki-67.
Several tumors had a level of Ki-67 labeling close to 0%, but tumor SUVs as high as >5 g/mL were observed (). This is in contrast to the studies in untreated tumors where there was generally very little [18
F]FLT uptake in tumors with a Ki-67 labeling index of less than ~20.9,31
One could speculate that this indicates DNA repair in the absence of proliferation, as recently reported.50
F]FLT uptake was observed in some tumors without measurable TK1 expression. In fact, the highest 2 SUVs (~5.1 and 6) were measured in lesions with no or very few TK1-positive cells. This might indicate that TK1 staining is insensitive to detecting active TK1. Alternatively, and despite our attempts to identify by light microscopy the tumor sections that showed greatest viability and mitotic activity, sampling errors cannot be ruled out with certainty. Regardless of the underlying mechanisms that cannot be uncovered from the current data, the lack of relationship between tissue markers of proliferation and tumor [18
F]FLT uptake after treatment raises questions about the usefulness of this approach in patients with soft tissue sarcoma.
This study has several limitations. First, measurement inaccuracies of TK1 and Ki-67 activity cannot be ruled out. However, the significant correlation between TK1 and Ki-67 staining index argues against such error. To accurately correlate [18F]FLT tumor uptake to immunohistochemical results, we conducted the imaging studies as closely as possible to the date of surgery (average time interval, 5 ± 2.9 days). Therefore, the time from end of treatment to follow-up imaging was more variable and averaged 13.6 ± 9.3 days.
Second, as described above, the study population was heterogeneous and included patients with various sarcoma subtypes, who underwent various treatments.
Third, the study population included only 20 patients with soft tissue sarcoma. Therefore, we cannot exclude a weak correlation between [18F]FLT uptake and Ki-67 labeling if more patients had been enrolled in this study. However, for [18F]FLT-PET to emerge as a clinically useful test for response assessments, a close correlation between [18F]FLT SUVs and histopathologic response and/or proliferation would be necessary. On the basis of our data, this appears to be very unlikely.
Fourth, we used SUVs to quantify [18
F]FLT uptake. This decision was based on a series of articles which demonstrated 1) that [18
F]FLT SUVs and influx constants (Ki
) show a close correlation in untreated tumors, 2) that [18
F]FLT SUVs correlate well with histopathologic markers of tumor cell proliferation, and 3) that [18
F]FLT SUVs are easier to measure clinically than tracer kinetic parameters. Further studies are needed to investigate whether tracer kinetic analysis51,52
F]FLT uptake in treated tumors allows more accurate assessments of tumor proliferation in soft tissue sarcoma than do SUV measurements.
Fifth, Eilber et al37
have shown that a large extent of tissue necrosis following neoadjuvant treatment of patients with sarcoma is associated with improved patient outcome. However, to the best of our knowledge, this has not been proven for patients with GISTs.
In summary, in patients with high-grade soft tissue sarcoma, [18
F]FLT PET/CT imaging does not reliably predict histopathological response to neoadjuvant therapy, and [18
F]FLT uptake is unrelated to TK1 and Ki-67 expression. These findings suggest that response assessments based on [18
F]FLT-PET analysis do not provide an advantage over [18
F]FDG-based response assessments in patients with soft tissue sarcoma. Moreover, the reasons for the uncoupling between [18
F]FLT retention and TK1 and Ki-67 activity in sarcoma warrants further studies. Given the limited value of [18
F]FLT-PET in predicting therapeutic responses in rectal cancer,23
germ cell tumors,27
and now in soft tissue sarcoma, further studies in other cancers are needed to better understand the altered [18
F]FLT kinetics in patients after treatment.