Background: AML patients that are treated with induction chemotherapy undergo an early bone marrow assessment approximately 10–14 days after beginning treatment to determine the response to therapy. This bone marrow assessment is an imperfect measure of disease response and lacks sensitivity and specificity in determining subsequent remission. A more accurate assessment of chemotherapy responsiveness performed early during induction chemotherapy could allow better prediction of later remission and earlier change in therapy to improve outcome among patients that are not responding well to treatment. Thymidine uptake can be used to assess chemotherapy sensitivity in vitro, and we hypothesized that this would also be true in vivo. We performed a pilot study of [18F] FLT PET/CT scans to predict response to induction chemotherapy among patients with AML.

Methods: Eight patients with AML were treated with induction chemotherapy and whole body [18F]FLT PET/CT scans were performed at different time points (pre-treatment, day 1, 3, 4, 5 and 14) using approximately 5 mCi of [18F]FLT. Two patients had scans before treatment and day 14. Ten adult subjects without hematologic disease also underwent [18F] FLT PET/CT and these studies were used as normal controls. The CT images were used to reconstruct a skeletal mask that was then used to extract the bone marrow [18F]FLT signal from the PET images. The scans from those with normal bone marrow were used to establish baseline assessment parameters. The images from the scans could be quantified by computer analysis into bone marrow standardized uptake values (SUV). The bone marrow SUV were compared between the AML patients (responders vs. non-responders) relative to normal controls. The mean SUV values (SUVmean), max SUV values (SUVmax) and coefficient of variation (CV) were used in the comparison.

Results: The PET/CT was well tolerated and no adverse events were noted, although one patient did not complete the scan due to the development of epistaxis. Of the seven patients who completed the scan, 3 patients entered a complete remission after a single course of induction (responders) and 4 patients either required two induction courses to achieve remission or had refractory disease (non-responders). The responders had recovery of an ANC > 500 by day 22, 28, and 35 suggesting the [18F]FLT did not delay bone marrow regeneration. The fall in the bone marrow SUV occurred quickly and as early as day 1 after the beginning of chemotherapy among responders. Both mean and maximum total bone marrow SUV was significantly lower in responders compared to non-responders: SUVmean: 0.76±0.04 vs. 1.23±0.22, SUVmax: 3.6±0.01 vs 7.7±2.2. In addition, it was noted that the variation in SUV throughout the skeleton was more uniform in responders than non-responders resulting in a difference in CV: 0.29±0.01 vs 0.54±0.10. The differences in SUV and signal variation in the bone marrow were easily apparent when viewing the images from the scans. The significant heterogeneity in axial skeletal uptake among non-responders was notable and suggests that chemotherapy responsiveness is not uniform throughout the bone marrow and that there likely are islands of persistent/unresponsive disease during therapy. It is not clear whether the variation is due to islands of drug resistance, differences in blood flow, or rates of response. This non-uniformity of marrow uptake among non-responders may explain the poor predictive value of the day 10–14 bone marrow aspirate drawn at the iliac crest.

Conclusion: In our small pilot study, [18F]FLT PET/CT could distinguish between responders and non-responders among patients undergoing induction chemotherapy for AML possibly as early as one day following the beginning of treatment. Also, the scans suggest that response to chemotherapy in AML does not occur uniformly throughout the bone marrow space and that in non-responders, there are islands of disease that persist after chemotherapy. These finding warrant further study of [18F]FLT PET/CT as a tool for early response assessment in AML.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author