**4.1 18F-FDG labeled leukocytes**

One of the earliest attempts at creating a more specific PET radiopharmaceutical for infection imaging was the development of an *in vitro* method for labeling autologous leukocytes with 18F-FDG [69, 70]. A recent meta-analysis indicates that 18F-FDG labeled leukocyte imaging accurately diagnoses infection [71]. Seven studies (*n* = 236) were included in the meta-analysis. Pooled sensitivity was 86.3% (95%CI: 75−92.9%) and pooled specificity was 92% (95% CI: 79.8−97.1%). The positive likelihood ratio was 6.6 (95% CI: 3.1−14.1) and the negative likelihood ratio was 0.2 (95% CI: 0.12−0.33).

In spite of these favorable results, 18F-FDG WBC has not been integrated into the routine diagnostic workup of infection. There are several reasons for this. Labeling efficiency is variable both in patients and normal volunteers, ranging from less than 25% to more than 95% [72–78]. This inconsistency makes it difficult to determine the quantity of 18F-FDG needed for labeling leukocytes. If a worst-case labeling efficiency scenario is assumed, that is, 35%, what happens if the labeling efficiency is 80%? Is the amount of activity reinfused is reduced accordingly? If so, will the number of labeled leukocytes reinfused be adequate to provide diagnostically useful data?

Stability of the 18F-FDG WBC label is another issue. In one investigation, leukocyte retention of 18F-FDG decreased from 39% to 44% at 90 minutes to 19% at 4 hours [75]. In an investigation of normal volunteers, mean leukocyte retention of 18F-FDG was 85% ± 4% at 1 hour, and 68% ± 7% at 4 hours [77]. In view of the degree of 18F-FDG elution, one has to question whether imaging findings reflect accumulation of 18F-FDG WBC, 18F-FDG, or a combination.

The 110 minute physical half-life of fluorine-18 is a significant disadvantage. The time needed for *in vitro* labeling, up to 3 hours, needs to be accounted for when determining the amount of activity used to label the leukocytes. The short half-life makes it impractical for labeling to be performed off-site, which is a significant limitation in the United States where the vast majority of these labelings are performed at outside radiopharmacies. In indolent, low-grade, infections, leukocyte accumulation is slow, and imaging at later time points (e.g., 24 hours) may be necessary. The short half-life of fluorine-18 precludes imaging more than 4–5 hours after reinfusion of labeled cells. For all of these reasons, it is unlikely that 18F-FDG-labeled leukocyte imaging will ever become part of mainstream clinical nuclear medicine.

#### **4.2 Copper-64 labeled leukocytes**

64Cu labeling of leukocytes also has been investigated. In 10 normal volunteers, the labeling efficiency, cell viability, and stability of 64Cu labeled leukocytes were compared with those of 111In labeled leukocytes and 18F-FDG labeled leukocytes [77]. The mean labeling efficiency for 64Cu labeled leukocytes, 87% ± 4%, was nearly identical to that of 111In labeled leukocytes 86% ± 4%. Leukocyte viability was the same for both radiolabels at 1 hour, 99% ± 1%, but was significantly higher for 64Cu labeled leukocytes than for 111In labeled leukocytes at 3 hours (98% vs. 96%, respectively) and at 24 hours (61% vs. 48%, respectively). Label stability was significantly higher for 111In labeled leukocytes at 1, 2, 3, 4, and 24 hours (94%, 93%, 92%, 91%, and 88%, respectively) than for 64Cu labeled leukocytes (91%, 89%, 88%, 86%, and 79%) and 18F-FDG WBC (85% ± 4%, 81% ± 4%,76% ± 4%, and 68% ± 7%). Unfortunately, the labeling procedure required the use of two chelating agents: tropolone to allow the 64Cu ion to enter the cell, and quin-MF/AM, to prevent elution. This complex, timeconsuming procedure, which requires skilled personnel, is not well suited to routine clinical use.

Chitosan nanoparticles also have been used to label human leukocytes with 64Cu. The labeling efficiency was only about 26% and more than 90% of the activity had eluted from the leukocytes at 2 hours [79].

### **4.3 Zirconium-89 labeled leukocytes**

89Zr, with a half-life of 78.4 hours, has also been used to label leukocytes *in vitro*. In one investigation, chitosan nanoparticles were used to label human leukocytes with 89Zr. Labeling efficiency was 76.8%. Cell viability at the completion of labeling was 61%; 28.4% of the intracellular activity had eluted at 2 hours, 35.2% at 4 hours, and 53.3% at 24 hours. The entire labeling process took nearly 6 hours to complete. In this investigation, only 61% of the labeled leukocytes were viable.

Recent investigations are more promising. In one study, *in vitro* labeling of human leukocytes with 89Zr-oxine was compared to labeling with 111In-oxine [80]. Labeling efficiency for 89Zr labeled leukocytes was 48.7% vs. 89.1% (*P* < 0.0001) for 111In labeled leukocytes. However, there were no significant differences between 89Zr labeled leukocytes and 111In labeled leukocytes with respect to elution of activity or cell viability. Another group obtained similar results when using 89Zr-oxinate4 to label human leukocytes [81]. These results are encouraging, but *in vivo* investigations of 89Zr labeled leukocytes to diagnose infection are lacking.
