Date Published: February 7, 2019
Publisher: Public Library of Science
Author(s): Mathieu N. Toledano, Pierre Vera, Hervé Tilly, Fabrice Jardin, Stéphanie Becker, Giorgio Treglia.
The study objective was to compare the prognostic value of interim and end-of-treatment FDG PET/CT using five therapeutic evaluation criteria in patients with diffuse large B cell lymphoma (DLBCL).
181 patients were retrospectively analysed. All patients underwent FDG-PET at baseline and after four cycles (iPET4) of first-line chemotherapy and 165 at the end-of-treatment (PET-eot). Ratio Deauville score (rDS) (SUVmax-target residual lesion/SUVmax-liver) was measured in iPET4 and PET-eot, and its optimal threshold was determined using receiver operating characteristic (ROC) curve analysis. Deauville score (DS) (iPET4 and PET-eot), ΔSUVmax, ΔSUVmax determined according to Menton 2011 criteria (ΔSUVmax+DS) and ΔSUVmax+rDS were also evaluated (iPET4 only). Median follow-up was 44 months.
ROC analysis revealed the optimal cut-off value was 1.4-fold of SUVmax-liver on iPET4 and PET-eot. On iPET4, positive predictive value (PPV) of rDS was significantly better than DS: 81.58% vs. 67.79%. In univariate analysis, the five interpretation methods were statistically significant (p<0.0001 for progression-free survival [PFS] and overall survival [OS]). In multivariate analysis, only rDS was an independent prognostic factor (p = 0.0002 and p<0.0001 for PFS and OS, respectively). On PET-eot, similarly, the two therapeutic evaluation criteria analysed (rDS and DS) were statistically significant at the univariate level (p<0.0001). rDS was the only significant prognostic factor in multivariate analysis (p<0.0001). PPV and accuracy of rDS were also better than DS. rDS with a tumor/liver ratio of 1.4 is a robust prognostic factor in patients with DLBCL on iPET4 and PET-eot.
Despite the improvement made by immunochemotherapy, 30% to 40% of patients diagnosed with large-B-cell lymphoma (DLBCL) will relapse , with a majority of patients dying of the disease . It is therefore crucial to identify these nonresponder patients to offer them new treatments. Fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) has been widely validated as a prognostic tool in DLBCL and Hodgkin lymphoma (HL) for both interim and end-of-treatment [3–7]. Interpretation criteria have evolved to decrease false positives, and the current recommendations are to use Deauville criteria (5-point scale-Deauville Score [DS]) with a positive PET defined by tumor residual uptake moderately higher than liver (score 4) or two to three times the liver SUVmax (score 5). These criteria are applicable for both interim and end-of-treatment PET . However, at present, changing treatment solely based on interim PET (iPET) is not recommended outside clinical trials . Indeed, a negative iPET is truly negative (e.g., high negative predictive value [NPV]) in most cases, whereas positive iPET may be truly or falsely positive, with a high percentage of patients with abnormal uptake presenting a good outcome . Studies using Deauville criteria analysis with a cut-off >3 for iPET in DLBCL reported good NPV with 2-year PFS rates between 84% and 85%, whereas the PPV ranges from 25% to 55% [9–11]. To improve iPET’s PPV, a quantitative approach was proposed after 2 and 4 courses of chemotherapy with a ΔSUVmax based on the reduction of tumor SUVmax from baseline. The ΔSUVmax method has shown to be more reproducible between readers and more robust than Deauville criteria by decreasing the number of false positives [4–7]. The PETAL study, including 600 patients with DLBCL, demonstrated that the ΔSUVmax approach with a cut-off of 66% of reduction (iPET2) was highly predictive of outcome . However, it does not appear effective for low-risk (International Prognostic Index [IPI] ≤1) patients, for whom the DS is accurate . In addition, the determination of ΔSUVmax has been modified according to Menton 2011 criteria , and these modified criteria were never validated (ΔSUVmax if eligible patient and DS if SUVmax initial tumor <10 and or if SUVmax residual tumor >5). Several authors suggested a quantitative adaptation of the DS to better separate a score 3 and 4 in HL [15, 16], DLBCL [17–19] and follicular lymphoma (FL) , using a ratio between the residual uptake and the liver. The aim of this study was to compare the prognostic value of these different therapeutic evaluation criteria after four courses of immunochemotherapy (iPET4) and at the end-of-treatment PET (PET-eot) in patients with DLBCL: ratio Deauville score (rDS), visual DS, ΔSUVmax, ΔSUVmax determined according to Menton 2011 criteria and ΔSUVmax + rDs.
The aim of our study was to evaluate the prognostic impact of the ratio between SUVmax of the hottest target residual lesion and SUVmax of the liver (rDS) in patients with DLBCL undergoing interim FDG-PET/CT (iPET4) during the first line of immunochemotherapy and PETeot and to compare rDS with criteria already validated i.e DS and ΔSUVmax on iPET4 and DS on PETeot. In our study, ROC analysis showed that the optimal cut-off for rDS was 1.4-fold of SUVmax liver on iPET4 and PET-eot. Prognosis accuracy using rDS has allowed a significant increase of PPV compared to Deauville criteria on iPET4 and PET-eot (81.58% vs. 67.79%, respectively, for PFS and 63.16% vs. 52.54%, respectively, for OS on iPET4; 94.12% vs. 75%, respectively, for PFS and 76.47% vs. 60.42%, respectively, for OS on PET-eot) and therefore decrease the number of false positives. The NPV of rDS remains satisfactory on iPET4 compared to DS (72.72% vs 75.41%, respectively, for PFS and 78.32% vs 80.32%, respectively, for OS) and comparable to DS on PET-eot (77.10% vs 77.77%, respectively, for PFS and 83.97% and 84.61%, respectively, for OS). This is the first study confirming the prognostic performance of Menton 2011 criteria, with an accuracy of 74.58% for PFS and 75.14% for OS. These results agree with the GAINED study, which shows the prognostic influence of iPET4 ΔSUVmax (using Menton 2011 criteria) in R-ACVBP and R-CHOP14 arms . After four cycles of chemotherapy, rDS, Menton 2011 criteria and ΔSUVmax had the best specificity, PPV and accuracy better than DS. But only rDS was found to be an independent prognostic factor on PFS and OS. ΔSUVmax was statistically significant at the multivariate level on PFS but not on OS. Prognosis performances of ΔSUVmax + rDS (iPET4) were between the three mentioned above, and DS and showed no particular benefit. NPVs were similar for all evaluation criteria on PFS and OS on iPET4.
To conclude, rDS with a tumor/liver ratio of 1.4 is a robust prognostic factor in patients with DLBCL on iPET4 and PET-eot and shows better prognosis accuracy and survival prediction than Deauville criteria.