Although BM aspiration with morphological assessment remains the standard for residual disease assessment in patients with AML, its predictive value for residual disease is suboptimal and the technique is invasive [5,6,7,8]. To complement this limitation of BM aspiration, noninvasive modalities have been applied to assess the presence of viable residual cells in BM after induction. Several studies have suggested that dynamic contrast-enhanced magnetic resonance imaging of BM can be used to assess the changes in microvascular density and could be a useful prognostic indicator of disease activity and survival [17,18,19]. 18F-FDG PET imaging, which measures glucose metabolism, has also been shown to have potential for evaluation of leukemic BM infiltration .
As a surrogate of cellular proliferation, 18F-FLT PET has been used for early therapeutic monitoring in various cancers [21,22,23,24]. The rate-limiting step for intracellular 18F-FLT retention in proliferating cells is phosphorylation by TK-1, and thus, 18F-FLT accumulates in proportion to TK-1 activity . A more than 10-fold overexpression of TK-1, which is the key enzyme for intracellular 18F-FLT accumulation, is observed in leukemic blasts . Only a few studies have reported the feasibility of 18F-FLT PET imaging in AML. In a pilot study by Buck et al. , patients with relapsed, refractory, or untreated leukemia showed higher 18F-FLT uptake in the BM and spleen than did controls. This indicated that 18F-FLT PET reflected disease activity, but there was no significant correlation between 18F-FLT uptake and the number of leukemic blasts identified in BM biopsy. The authors discussed that the proliferative activity of leukemic blasts and normal BM cannot be differentiated using 18F-FLT PET, as the normal hematopoietic cells also show increased 18F-FLT uptake. In another study of AML patients who underwent induction therapy, the BM uptake in RD patients was significantly greater than in normal controls, while the BM uptake in CR patients was similar to that in normal controls in pre-treatment 18F-FLT PET images . That study also investigated whether 18F-FLT PET during or after induction therapy could be useful for early assessment of the clinical response. All 5 CR patients exhibited low BM uptake, while 2 RD patients displayed elevated BM uptake (maximum SUV, 3.6 ± 0.4 vs. 11.4 ± 0.8, p < 0.001). Thus, the study concluded that 18F-FLT PET during induction chemotherapy could predict successful BM ablation early on. We presumed that increased 18F-FLT uptake is associated with the degree of RD in the present study, because all PET/CT scans were performed shortly after myeloablative chemotherapy. Our results indicated that interim 18F-FLT PET/CT performed well in the early prediction of clinical response to induction therapy in patients with AML. All 5 patients with PET-negative findings achieved CR in the follow-up BM aspiration 4−6 weeks after induction therapy (high negative-predictive value), and both of the RD patients showed PET-positive findings (high sensitivity).
Leukemic distribution can be heterogeneous or localized, despite the malignant systemic disease [28, 29]. Vanderhoek et al.  demonstrated substantial heterogeneity of BM uptake in the 18F-FLT PET response assessment, both during and after induction therapy. The distribution of 18F-FLT uptake in RD patients was more heterogeneous (higher CV) than that in CR patients. Additionally, 1 RD patient was in an aplastic state at the interim BM biopsy, despite the significantly increased BM uptake in the 18F-FLT PET image. The BM biopsy is usually performed on the unilateral posterior iliac bone and the BM evaluation is thus limited to that area and cannot reflect the whole of the BM in vivo; therefore, residual leukemia might be missed. 18F-FLT PET imaging is performed from the vertex to the upper thigh after a single injection, facilitating noninvasive assessment of the total hematopoietic BM compartment, which represents a significant advantage over a BM biopsy.
The present study also suggested that 18F-FLT PET/CT may perform better than follow-up BM assessment in predicting relapse. Three patients with PET-positive findings achieved CR in the follow-up BM aspiration (false-positive). Of these, 2 patients (patients 6 and 7) had confirmed or clinically suspected relapse within 4 months after consolidation, even though 1 patient had been classified as favorable-risk. On the other hand, patient 3, with relapsed AML, had PET-negative findings and has now been in the third remission for 34 months, although relapse after achieving remission is considered a poor prognostic factor in patients with AML . One of the 5 PET-negative patients relapsed; this patient showed a heterogeneous BM distribution (CV = 0.32). In a study conducted by Vanderhoek et al. , the CVs of all patients with residual/refractory disease on the interim BM biopsy were ≥ 3.0. BM heterogeneity as well as visual analysis may be important for predicting relapse. Current risk-adapted therapy in adult patients with AML is based on only a few prognostic markers, such as age, cytogenetic risk, and gene mutations at diagnosis. More recently, adjustment of therapy based on incorporation of post-treatment data is likely to become increasingly important. When added to the current risk-stratification system, 18F-FLT PET/CT may provide a more accurate selection of patients at high risk for relapse, and thus improve survival by modification of treatment.
In our CR patients, 1 false-positive PET case (patient 9) showed persistent mild thrombocytopenia (Table 6). Agool et al.  reported that significantly increased 18F-FLT uptake was observed in patients with myelodysplasia, myeloproliferative disorder, and myelofibrosis. Although our patient was not further evaluated for thrombocytopenia, there was a possibility that BM dysfunction caused the false-positive PET finding.
The main limitation of this study is the small sample size. Additionally, molecular studies were not available in all patients. Risk stratification, based on combined clinical and molecular markers, has recently been proposed to improve the predictive value of early response assessment [1, 4]. Our results suggest that further studies on the prognostic potential of the combination of molecular abnormalities and interim 18F-FLT PET/CT results are warranted. The heterogeneity of compliance to the imaging protocol is another potential limitation. We originally planned to standardize the imaging protocol strictly. However, variability in the injected activity and in the time from injection to PET scanning was unavoidable, and some cases had accidental extravasation at the injection site. Poor protocol compliance may lead to over- or underestimation, particularly in terms of quantitative PET analysis. To overcome this limitation, we performed both qualitative and quantitative analyses. In qualitative analysis, we experienced no difficult cases when determining whether the image was positive or negative according to our criteria. For solid cancers that are already frequently and usefully evaluated with PET/CT, qualitative PET analysis is as important as quantitative PET analysis. For lymphoma, the Deauville criteria is based on visual analysis . Lastly, there were neither standards in place regarding the time for the 18F-FLT PET/CT imaging of patients with AML nor any established criteria for interpreting the images. The role of imaging studies, including PET/CT, is still limited in terms of evaluation of hematologic malignancy. However, application of 18F-FLT PET/CT to patients with AML has been continuously studied although the number is small [15, 27, 32]. Recently, a phase II clinical trial assessing 18F-FLT PET/CT in AML was launched , and this prospective ECOG-ACRIN EAI141 is likely to provide more conclusive results when completed. Our preliminary results could be of value in planning future clinical studies (e.g., determining appropriate time points for performing 18F-FLT PET/CT), data analysis (e.g., setting the criteria for visual analysis), and the clinical interpretation of 18F-FLT PET/CT images in patients with AML. We anticipate that these study results will collectively contribute to the establishment of clinical applications of 18F-FLT PET/CT imaging in patients with AML.