Dear Sir,

In their interesting article, Hasenclever et al. [1] proposed a quantitative extension of the Deauville scale to a continuous scale for assessing response in interim FDG PET scans performed in lymphoma patients. In an individual, they defined “qPET” as the ratio of SUVpeak (involving SUVmax+ 3 hottest adjacent voxels) assessed in a target lesion to a reference liver SUVmean obtained from a 30-ml cuboid volume. The authors concluded that qPET<1.3 excludes abnormal response with high sensitivity.

Although the qPET scale, like the Deauville scale, minimizes the influence of some factors either physiological (i.e. blood glucose level) or technical on its variability, the relative measurement error (MEr) should nevertheless be taken into account. MEr is the relative difference between a single estimate of a parameter and its average “true” value [2] and applies both to SUVpeak and SUVmean when a qPET value in an individual is to be compared to a cut-off value: ΔqPET/qPET = ΔSUVpeak/SUVpeak + ΔSUVmean/SUVmean.

We have recently published the variability of average SUV obtained by pooling n = 5–10–15–20–25–30 hottest voxels [3]. MEr of SUVpeak in a target lesion can be estimated to be 10.8–14.2 % (n = 5; 95–99 % reliability, respectively). MEr of SUVmean in the liver can be estimated from Boktor et al.’s results in lymphoma patients [4]: 31.1–40.9 % (=100*[0.50/(2.23*21/2)]*1.96–2.58, with 95–99 % reliability, respectively [2]). Therefore, MEr of qPET may reach 41.9–55.1 % (95–99 % reliability, respectively). This result should be compared to the relative difference between the cut-off values proposed by Hasenclever et al. [1]: between 1.3 and 0.95 and between 1.3 and 2.0, this difference is 26.9 and 53.8 %, respectively. In other words, when applied to a cut-off value of 1.3, qPET MEr may be greater than the relative difference between 1.3 and 0.95 with 95 % reliability and even greater than that between 1.3 and 2.0 with 99 % reliability, corresponding to Deauville categories 3–4 and 4–5, respectively.

The above calculated estimation indicates that SUV MEr may impair qPET methodology. We therefore suggest that qPET variability should be specifically investigated in a series of lymphoma patients. Indeed, to the very best of our knowledge, it is not known whether, in an individual, target SUVpeak and liver SUVmean are actually independent. Moreover, SUVpeak values in Laffon et al. [3] are larger than those usually assessed in interim FDG PET scans in lymphoma leading here to SUVpeak MEr underestimation.