Recently, Bailly et al. reported results from the LyMa-PET project regarding the prognostic value of image-derived F-fluoro-deoxyglucose positron emission-tomography (F-FDG-PET) quantitative indices in patients with mantle cell lymphoma (MCL) having autologous stem-cell transplantation (ASCT).1 The authors concluded that the maximal standard uptake value (SUVmax) corresponding to the hottest voxel of the lesion with the highest uptake at diagnosis, has a strong prognostic value for both progression-free survival and overall survival. A new scoring system combining the mantle cell lymphoma international prognostic index (MIPI) score and SUVmax was proposed to improve the patient outcome prediction. In particular, the authors argued that, since the prognostic value of SUVmax and SUVpeak (corresponding to the average SUV obtained from a 1-mL sphere centered over the most active region of the highest-uptake lesion) was similar, the former metric was preferred over the latter one owing to its wide use.
We suggest that, for such a choice, the comparison between the measurement uncertainty (MU) of SUVmax and SUVpeak should be taken into account and discussed in details. Indeed, SUVpeak, or any averaged SUV from several voxels such as the SUVmax-N that pool several N hottest voxels regardless of their location within one or different F-FDG-positive lesion, have a lower MU than the former, and, hence, are more reliable for a clinical decision to be taken.32 In a previously published lung cancer series, the relative measurement error (MEr) (i.e., the relative difference between a single SUV estimate and its average true value) of SUVpeak and SUVmax-40 was found to be significantly lower than that of the SUVmax: 9.4 and 8.8 versus 13.9 % (with 95 % reliability), respectively.3 These results may be applied to MCL patients because positron emission tomography (PET) imaging does not allow identifying the disease type underlying an F-FDG uptake. Noteworthy, they were obtained with SUVmax values ranging between 6.6 and 23.2 g/mL, and it should be stressed that the MEr percentage does increase when applied to SUV values lower than 6.6 g/mL, as clearly demonstrated by de Langen (in terms of repeatability percentage): the lower the SUV value, the greater its MU.4 In the study by Bailly et al. Online Supplementary Table S2 reports a minimal value at baseline of 1.8 and zero g/mL for SUVmax and SUVpeak, respectively, whose MEr is substantial, even incalculable.1 As a consequence, the inclusion of such low SUV outcomes may very likely explain the huge range of ΔSUVmax and ΔSUVpeak outcomes between baseline and before transplantation that are reported in Online Supplementary Table S4: [−100,+271] and [−95,+278] (expressed in %), respectively.1 Furthermore, the new scoring system combining the MIPI score and SUVmax that was proposed by the authors, used a SUVmax cutoff value of 10.3 g/mL. Since MEr is the relative difference between a single estimate and its average true value, we suggest that the cutoff value of 10.3 g/mL be completed by lower and upper limits of 8.8 and 11.7 g/mL obtained from the above-reported ± 13.9 % MEr for SUVmax (95 % reliability). The use of such limits, which may be refined by taking into account the MU of the cutoff outcome and the PET system employed, could enable physicians to adjust their decision according to the clinical trial design, that is, to avoid a false-negative/positive PET scan leading to patient’s under-treatment/therapy-escalation, respectively. We believe that this point emphasizes why a metric with a reduced MU should be preferred. Finally, it is worth noting that the MU argument also applies to the therapeutic evaluation since the repeatability, i.e., the minimal relative change between two SUVs assessed from two successive examinations that is required to consider a significant difference, can be computed as 2×MEr.5
To conclude, the relevant study by Bailly et al. clarifies the prognostic value of quantitative F-FDG-PET imaging in MCL patients. With respect to establishing the prognosis or to assessing the response to treatment, we suggest that the criteria for the choice of a metric should involve the magnitude of its MU and, therefore, averaged quantitative indices should be a priori preferred. How acceptable the MU magnitude of a metric should be, that might cause difficulties in clinical decision making and, hence, that would rule it out, is a question of judgment and consensus.
- Bailly C, Carlier T, Berriolo-Riedinger A. Prognostic value of FDG-PET in patients with mantle cell lymphoma: results from the LyMa-PET Project. Haematologica. 2019;1. Google Scholar
- Wahl RL, Jacene H, Kasamon Y, Lodge MA. From RECIST to PER-CIST: evolving considerations for PET response criteria in solid tumours. J Nucl Med. 2009; 50(Suppl 1):122S-150S. PubMedhttps://doi.org/10.2967/jnumed.108.057307Google Scholar
- Laffon E, Burger IA, Lamare F, de Clermont H, Marthan R. J Nucl Med. 2016; 57(1):85-88. PubMedhttps://doi.org/10.2967/jnumed.115.161968Google Scholar
- de Langen AJ, Vincent A, Velasquez LM. Repeatability of 18F-FDG uptake measurements in tumors: a meta-analysis. J Nucl Med. 2012; 53(5):701-708. PubMedhttps://doi.org/10.2967/jnumed.111.095299Google Scholar
- 2008. Google Scholar