Assessment of response by RECIL 2017 versus Lugano 2014 criteria in patients with previously untreated CD20-positive DLBCL

Response assessment in lymphoma has advanced over recent years following the development and utilization of more advanced imaging techniques and evaluation methods. The previously accepted standard criteria developed by the International Working Group, 1999, was superseded by the Cheson criteria in 2007, which incorporated the use of positron-emission tomography (PET), immunohistochemistry and flow cytometry into the assessment.^1,2 Subsequently, in the Lugano 2014 guidelines, the use of ¹⁸F fluorodeoxyglucose (FDG) PET-computer tomography (CT) was formally incorporated into standard staging for FDG-avid lymphomas. For non-FDG avid lymphomas, the Lugano 2014 criteria uses bi-dimensional tumor measurements for up to six CT target lesions if PET is unavailable.³ More recently, in 2017, Anas Younes et al., published the Response Evaluation Criteria in Lymphoma (RECIL) criteria which hypothesized that uni-dimensional measurements of up to three target lesions could be equally as accurate as the Lugano 2014 criteria.⁴

In order to develop the RECIL 2017 criteria, Anas Younes et al., analyzed 47,828 imaging measurements of 2,983 patients with lymphoma who were enrolled on 10 multicenter clinical trials. They demonstrated that using the sum of the longest diameters of a maximum of three target lesions that it is possible to assess tumor burden in patients with lymphoma at a similar level to that of the current standard criteria (Lugano 2014).⁴ However, no study has yet compared the Lugano 2014 criteria and RECIL 2017 criteria. This current analysis, presented during the 61^st American Society of Hematology (ASH) Meeting & Exposition by Lale Kostakoglu, Icahn School of Medicine at Mount Sinai, New York, US, sought to compare response assessment using the Lugano 2014 and RECIL 2017 criteria in patients enrolled on the phase III GOYA study.⁵

GOYA study

The phase III GOYA study (NCT01287741) compared the efficacy of obinutuzumab (G) to rituximab (R) in combination with cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP; G-CHOP, R-CHOP) in previously untreated patients (n= 1,418) with CD20-positive diffuse large B-cell lymphoma (DLBCL). Patients were randomized 1:1 to either G-CHOP or R-CHOP and response assessment by FDG-PET was conducted at screening and 6–8 weeks after last study treatment.

The original prospective analysis of response in the GOYA trial was conducted using the Cheson 2007 criteria, with a subsequent retrospective analysis using the Lugano 2014 criteria. This additional analysis retrospectively evaluated responses by the RECIL 2017 criteria and assessed whether they were comparable to the Lugano 2014 evaluation. In total 1,334 patients from the GOYA study had available PET data and were included in the evaluation.

Read more about the GOYA trial on the Lymphoma Hub here.

Lugano 2014 vs RECIL 2017⁵

The main differences between the Lugano 2014 and RECIL 2017 assessment criteria are shown in Table 1.

• Concordance between Lugano 2014 and RECIL 2017 criteria for end-of-treatment (EoT) response is shown in Table 2 below
  • There was a good correlation between RECIL 2017 and Lugano 2014 criteria for complete response (CR):
    • 92.5% (894/966) of patients in complete metabolic response (CMR) at EoT by Lugano 2014 were classified as being in CR by RECIL 2017
  • However, there was a poor agreement between the classification of patients with progressive metabolic disease (PMD) at EoT by Lugano 2014 criteria and partial response (PR) or minor response (MR) by RECIL 2017 criteria: 63.5% (40/63)
  • Of the 43 patients with PMD by Lugano 2014 and non-PD by RECIL 2017, at EoT, 15 patients had no progression-free survival (PFS) event by Lugano as verified by CT by Cheson 2007 criteria, and could therefore be considered false positives

Subsequently, landmark PFS analysis by EoT response was conducted which found EoT CR status by RECIL 2017 criteria was highly prognostic for PFS (stratified hazard ratio [HR]= 0.35, 95% CI, 0.26–0.46, p< 0.0001) and overall survival (OS) (stratified HR= 0.26, 95% CI, 0.19–0.36, p< 0.0001).

• Multivariate analysis confirmed EoT CR status by RECIL 2017 criteria was prognostic for:
  • PFS: HR= 0.18, 95% CI, 0.15–0.23, p< 0.0001
  • OS: HR= 0.23, 95% CI, 0.17–0.32, p< 0.0001
  • This was independent of stratification risk factors such as international prognostic index score and geographical region

Assessment of PFS using RECIL 2017 criteria also provided similar results to the GOYA study:

• RECIL 2017: HR= 0.95, 95% CI, 0.78–1.17, p= 0.64
• GOYA: HR= 0.94, 95% CI, 0.78–1.12, p= 0.47

Cheson 2007 vs RECIL 2017⁵

Since the original analysis of GOYA used the Cheson 2007 criteria, the authors of this study also compared the Cheson 2007 response assessment to RECIL 2017:

• Comparison of PFS between treatment arms (G-CHOP vs R-CHOP):
  • Cheson 2007: HR= 0.94, 95% CI, 0.78–1.12
  • RECIL 2017: HR= 0.98, 95% CI, 0.80–1.21
• Three-year PFS rates were slightly lower with Cheson 2007 criteria compared with RECIL 2017:
  • Cheson 2007 (G-CHOP vs R-CHOP): 71.4% vs 69.8%
  • RECIL 2017 (G-CHOP vs R-CHOP): 75.0% vs 74.8%

Conclusion⁵

In this retrospective analysis, EoT CR status by RECIL 2017 criteria was found to be correlated to CMR status by Lugano 2014 criteria. Additionally, in the GOYA study, EoT CR status was prognostic for PFS in patients with previously untreated DLBCL. Taken together, these results indicate that RECIL 2017 criteria could provide a more simplified way of response assessment and PFS rate compared with the Lugano 2014 criteria.