Page 224 - 18F-FDG PET as biomarker in aggressive lymphoma; technical and clinical validation

Chapter 8
robust analysis of effect modifiers. We were also able to correct for differences in baseline patient characteristics between individual patients. Moreover, there was low risk of bias in our included studies according to QUIPS screening criteria. We decided to truncate survival at 2 years, because most clinically relevant events occur during this period. A recent IPD analysis showed that patients who are alive without progression at 2 years have survival rates similar to those of the age-, sex-, and country-matched population 7 years after this time [27].
Table 2. PPV and NPV using DS4-5, DS5, or ∆SUVmax to assign a PET-positive result at I-PET2 and I-PET4.
I-PET criteria
I-PET2 DS1-3 vs DS4-5 DS1-4 vs DS5
∆SUVmax I-PET4 DS1-3 vs DS4-5
DS1-4 vs DS5 ∆SUVmax
PPV (95% CI)
30.5 (26.2-33.8) 68.5 (56.6-80.3) 45.7 (37.3-54.1) 42.6 (33.0-52.3) 70.0 (51.7-88.3) 57.4 (41.2-73.5)
NPV (95% CI)
82.9 (80.0-85.7) 80.0 (77.5-82.5) 80.6 (78.1-83.2) 84.7 (81.1-88.3) 81.5 (77.9-85.0) 82.2 (78.0-86.4)
A limitation of this study was that for some patients, the baseline PET scan was not performed, which precluded calculation of ∆SUVmax. However, this should not bias our results because the DS was not different between the patients with and without a baseline PET/CT scan. We decided to use PFS as the primary outcome parameter because it is widely accepted. However, PFS is affected by age [28]. Outcome of older patients is determined not only by lymphoma but also by age-related comorbidities, adverse treatment effects, and limited life expectancy in general. Note that all findings were consistent when considering TTP and OS instead of PFS.
Future studies should focus on improving the PPV by further stratifying patients into risk groups based on baseline PET characteristics such as metabolic tumor volume [29] and dissemination [30] and by improving the criteria for assigning a PET-positive result at I-PET. In addition, the effect of therapy on I-PET criteria requires further study, because all patients in our analysis were treated with R-CHOP.
In conclusion, the best response criterion at I-PET was ∆SUVmax, which had higher discriminative power and predictive values than DS4-5 criteria. Although
222