In this population of patients treated with CD19 CART, both the established and explorative lymphoma response criteria showed considerable discordance in imaging endpoints based on different reasons for definition of PD. While the ORR was almost unaffected, classification of patients as SD and PD differed significantly. In addition, some patients with the new proposed response categories MR by RECIL and IR by LYRIC were identified. Dichotomization into responding and non-responding patients based on 3-month FU stratified OS by all criteria. Interestingly, grouping patients based on the Lugano TL PD into groups with uni TL, oligo TL and multi TL progression showed a significant trend for OS stratification.

Lymphoma response criteria have historically been developed and established in the first-line treatment setting, and notably in an era of cytotoxic chemotherapies. The Cheson and Lugano criteria have evolved from the unidimensional Response Evaluation Criteria in Solid Tumors (RECIST1.1) criteria [16] and measure the bidimensional extension for this typically nodal-dominant tumor phenotype [7; 9]. On CT imaging, patients with pretreated lymphoma often have residual masses that can easily be mistaken as vital tumor [17]. Therefore, the imaging response criteria for lymphoma incorporate metabolic activity of lymphoma manifestations as visualized by 18 F-FDG PET/CT in order to identify a complete response [7; 9,10,11].

In the development of the response evaluation criteria in lymphoma (RECIL), the panelists set out to facilitate the response assessment of lymphoma by reducing the number of target lesions that need to be measured to achieve similar validity as the Lugano criteria [11]. A reduction to 3 target lesions to represent tumor burden enabled robust response classification compared to the other criteria that rely on 6 manifestations. In addition, RECIL require a combination of depth of response and reduction of metabolic activity to classify a response.

In the first-line treatment setting, the association of imaging endpoint surrogates of survival as PFS is known to be strong regarding OS [18; 19]. However, in the course of the disease with r/r lymphomas, changes in the phenotype and metabolism of the manifestations occur. Typically, distant lymph nodes are more commonly affected and extranodal lesions are more frequent. In addition, preexisting remnants may be mistaken for active lymphoma if previous imaging studies are not reviewed. Notably, response criteria have not been adjusted for such changes in disease progression, neither in solid nor hematologic malignancies, and data on the association of PFS and OS in lymphoma is scarce [7; 16].

In the context of CAR T-cell therapy data on characterization of response by classification system is very limited. A single-arm, prospective study of 7 patients with LBCL and FL, treated with CD19 CART evaluated early response according to Lugano criteria on 1-month 18 F-FDG PET/CT [20]. Interestingly all patients in this study with less than CR subsequently relapsed [20]. A multicenter study with 171 patients analyzed the Deauville score of NHL patients under CART in 1 month FU 18 F-FDG PET/CT with similar results. Patients with Deauville Score 1 + 2 at 1 month FU had an improved long-term outcome compared to patients with Deauville Score 3–5, who were at risk for an early relapse. Moreover, all patients with Deauville Score 5 relapsed by month 3 [21]. Another group suggested a SUVmax ≥ 10 at 1 month as a predictor of progression [22]. Recently, it was shown that pretreatment tumor burden metrics of lymphoma under CART vary significantly based on the assessment method, impacting their association with survival outcomes [23].

The analysis of specific response patterns of lymphoma and the impact of pseudoprogression in the context of CART, has not yet been studied in detail [24]. Pseudoprogression is defined by a transient increase in tumor size due to an infiltration of the tumor by immune cells and is mainly described in solid tumors under immunotherapy, particularly in melanoma, affecting 5–12% of cases [25]. Few studies described cases of pseudoprogression after CART analogous to solid tumors [26].

To prevent patients with pseudoprogression from being misclassified as progressive disease LYRIC introduced the category of indeterminate response (IR) with 3 subcategories: IR1, increase in SPD increase ≥ 50% within the first 12 weeks of therapy, without clinical deterioration; IR2, appearance of new lesions, or growth of one or more existing lesions ≥ 50% at any time during treatment in the absence of overall progression; IR3, increase in FDG uptake of one or more lesions without a concomitant increase in lesion size or number [10]. LYRIC suggests follow-up in all IR cases after 12 weeks and encourages a biopsy for IR1 and IR2. In contrast to LYRIC, the other response criteria do not provide recommendations for lesion follow-up [7; 9; 11].

Another feasible method for distinguishing pseudoprogression from true progression would be immuno-PET, which uses mAbs or antibody fragments radiolabeled with a positron emitter radionuclide that can be detected on PET/CT imaging [27]. For lymphoma patients, there are a variety of potential targets, such as T-cell markers (CD3, CD4, and CD8), B-cell markers (CD19 or CD20), and immune checkpoints (PD-1, PD-L1, or CTLA-4) [28]. The first clinical trial with 5 patients included demonstrated the suitability of immuno-PET with 89Zr-rituximab (anti-CD20 MAC) in patients with relapsed B-cell NHL [29]. In a later study, iPET with 89Zr-labeled anti-CD20 mAbs was suggested as a potential biomarker for predicting the response of r/r DLBCL [30]. Immuno-PET has also been used to visualize the migration, activation, and expansion of CD19-specific CAR-T cells in an in vivo mouse model of B-cell lymphoma [31]. However, there is very few clinical data, especially in the context of CART and limited availability of immune-PET in clinical routine.

In our study, there was no difference in terms of OS in patients with NL PD, even when sub-analyzed by the location of newly appearing lesions. In addition, patients with LYRIC-based IR had a nonsignificant difference in survival compared with patients with PD. Therefore, patients with solely newly appearing lesions and patients with LYRIC-based IR should be further investigated regarding clinical benefit and may represent a new response category. NL biopsy for histological workup should be considered in these cases. Alternatively, liquid biopsy using ctDNA may represent a minimally invasive test to resolve diagnostic uncertainties in this clinical scenario [32; 33].

In addition, LYRIC-based PFS showed the strongest association with OS. One explanation could be that LYRIC effectively classifies patients with pseudoprogression into one of the IR categories. Another explanation is that patients with a small increase in tumor burden or growth of a single TL are also classified as IR, in contrast to the Lugano or Cheson criteria, in which a single significantly growing TL is classified as PD. In these patients, the lymphoma may indeed progress, but perhaps with low kinetics, resulting in longer OS. This would be consistent with our findings that patients with single-site TL PD have a longer OS than patients with oligo- or multi-site TL PD. To address this issue, further characterization of tumor kinetics would be interesting. Recently, it has been shown that the increase in tumor growth rate post-baseline compared to pre-baseline in lymphomas treated with CART has a significant impact on OS [34].

Future response assessment in lymphoma with novel imaging endpoints and response criteria will likely be based on assessment of whole tumor burden (e.g. metabolic tumor volume) and not only based on selected lesions. In the first-line setting of LBCL, the recently published International Metabolic Prognostic Index (IMPI), that consists of metabolic tumor volume, age, and stage, has outperformed the conventional IPI in estimating outcome [35]. These imaging findings may be integrated with prognostic risk-stratification tools such as the CAR-HEMATOTOX [36; 37]. Further areas of study may also focus on patterns of response, e.g. volume changes or the absolute number and size of new lesions. Such criteria refinements have been successfully applied in other cancer entities in the advanced, later-line disease stage, for example in metastatic prostate cancer [38; 39].

Recently, differences in imaging endpoints among response criteria in lymphoma were reported [14]. To our knowledge, there is no further literature comparing the response assessment in the context of r/r lymphoma under CART. Our study has limitations which need to be considered when interpreting the results. First, this is a single-center study with a limited number of subjects. Second, there were a few patients that were missed to follow-up or had no measurable disease. Not all patients had a PET-based assessment at day 30.