*3.3. Laboratory Parameters*

In the group of patients achieving CR after CAR-T cell therapy, the mean serum LDH proved significantly lower compared to the PR group at baseline (240 ± 28 U/L [CR] vs. 443 ± 262 U/L [PR]; *p* < 0.05) (Figure 4a). The CRP levels in serum at baseline also proved significantly lower in the CR group compared to the PR group (2.6 ± 4.6 μg/dL [CR] vs. 4.2 ± 6.6 μg/dL [PR]; *p* < 0.05) (Figure 4b). At baseline, no differences were found for the leucocyte count between patients achieving CR (5127 ± 766 1/μL) and PR (5290 ± 3142 1/μL; *p* > 0.05).

**Figure 4.** (**a**) Serum LDH at baseline before CAR-T cell therapy in patients achieving CR (240 ± 28 U/L) vs. patients achieving PR (443 ± 262 U/L; *p* < 0.05). (**b**) CRP at baseline before CAR-T cell therapy in patients achieving CR (2.6 ± 4.6 μg/dL) vs. patients achieving PR (4.2 ± 6.6 μg/dL; *p* < 0.05). The asterisk (\*) indicates clinical significance (*p* < 0.05).

In the whole patient cohort, we observed a significant decrease of CRP (2.8 ± 5.6 μg/dL vs. 0.1 ± 0.2 μg/dL; *p* < 0.01) and leucocyte count (5348 ± 2514 1/μL vs. 2954 ± 2024 1/μL; *p* < 0.01) and trend towards lower serum LDH (366 ± 235 U/L vs. 242 ± 69 U/L; *p* > 0.05) after CAR-T cell therapy.

The mean time interval between re-infusion of CAR-T cells and IL-6 peak was 6.5 ± 4.3 days and between re-infusion of CAR-T cells and Il-2R peak the time was 7.4 ± 4.1 days. The IL-6 peak (867 ± 951 ng/L vs. 9121 ± 11,266 ng/L) and IL-2R peak (2483 ± 1164 U/mL vs. 5548 ± 3949 U/mL) were lower in patients with CR compared to PR, without statistical significance (*p* > 0.05) (Figure 5).

**Figure 5.** IL-6 peak (867 ± 951 ng/L vs. 9121 ± 11,266 ng/L) and IL-2R peak (2483 ± 1164 U/mL vs. 5548 ± 3949 U/mL) in patients with CR vs. PR (*p* > 0.05).

After CAR-T cell therapy, 20/21 patients developed a cytokine release syndrome (CRS). Of these, 10/21 patients had a CRS grade I, 7/21 patients had a CRS grade II, 2/21 patients had a CRS grade III and 1/21 patient had a CRS grade IV. In the group of patients with CR after CAR-T cell therapy, two patients had a manifest CRS (grade > 1), whereas eight patients with PR developed a manifest CRS.

## **4. Discussion**

In this study, we investigated the predictive value of textural features derived from CT and volume-based parameters derived from 18F-FDG-PET/CT in comparison to serologic markers in patients with DLBCL undergoing CD19-CAR-T cell therapy.

Patients achieving PR had a considerably higher MTV and TLG in the baseline setting compared to patients achieving morphological and metabolic CR. This is not surprising, as the MTV was already found to have predictive value in lymphoma [14–16]. Xie et al. reported a negative progression-free survival (PFS) in patients with DLBCL presenting high MTV and SUVmax values in the pre-treatment setting, irrespective of the applied treatment regimen [14]. Similarly, Zhou et al. described MTV as the only independent predictor of progression-free survival and overall survival in their patient cohort with DLBCL undergoing R-CHOP therapy [15]. In a subsequent analysis, the same authors found baseline TLG to be a significant predictor of PFS [16]. Quantification of the volumetric parameters MTV and TLG and calculation of cut-off values for response prediction were also performed by Xie and Sasanelli [16,17]. Furthermore, response prediction based on MTV was performed in patients with follicular lymphoma undergoing immuno-chemotherapy [18].

Expectedly, quantitative 18F-FDG-PET proved also to be a good therapy monitoring tool with significant reduction in glucose metabolism in patients achieving PR and no residual uptake in patients achieving metabolic CR.

Other quantitative predicting and response monitoring metric features are those derived from texture analysis of either PET-metabolic or CT-morphologic image data, which are then post-processed by using radiomics parameters [13]. At this point, our

results obtained from CT-radiomics analysis indicate that with increasing tumor tissue homogeneity and correspondingly decreasing heterogeneity in the baseline setting, the chances of achieving CR significantly increase. Similar results were reported in DLBCL patients undergoing immunochemotherapy [10]. Aide et al. demonstrated that 18F-FDG-PET heterogeneity of the largest lymphoma lesion was an independent predictor of two years-event-free survival [10]. In this study, the only independent predictor when analyzed together with IPI and MTV was the long-zone high-grade level emphasis.

Texture analysis was further demonstrated to have potential in improving the value of pretreatment PET/CT for prediction of the interim response of primary gastrointestinal-DLBCL [9]. In this report, the SUVmax, MTV, as well as the entropy were significantly higher in the non-CR group. In the report by Aide et al., texture analysis of the skeleton in patients with DLBCL proved beneficial for diagnosis of infiltration with skewness being the only independent predictor of PFS [19].

The assumption that images contain information of disease-specific processes is the basis for the use of radiomics, which aims at enhancing the existing data by means of advanced mathematical analysis. Various studies from different fields in imaging highlight the potential of radiomics to support clinical decision-making [20,21]. However, various technical factors may influence the extracted radiomic features, which is a limitation of this approach that needs to be considered [22].

Although radiomics applications have yet to arrive in routine clinical practice, image interpretation using radiomics has potential in terms of a more personalized medicine in the future.

Another discipline in which radiomics is now increasingly being applied is pathology [23]. The idea to genetically classify tumors without biopsy using non-invasive extraction of image information promises support in diagnostics, individualized prognosis and therapy planning.

LDH is a non-specific marker for lymphoma whose prognostic significance is well established for both indolent and aggressive lymphomas at the time of diagnosis, which indirectly reflects the tumor burden. As expected, we found lower serum levels of LDH at baseline in DLBCL patients, which reached CR after CAR-T cell therapy. LDH significantly decreased after therapy accompanied also by decreasing serum levels of CRP and leucocyte count.

Of interest, the IL-6 and IL-2R serum peaks measured in the first week after CAR-T cell therapy onset proved lower in patients classified CR compared to patients classified PR. Almost half of our patient cohort developed a manifest CRS (grade 2 or higher) after CAR-T cell therapy, which was mainly observed in patients with PR after treatment.

Our results are in line with already existing data with respect to response prediction in DLBCL showing similar results of studies exploring the role of PET/CT and CT in immunotherapy.

Our study has some limitations. First, not all patients received the same CD19-CAR-T cells compound. Second, due to the retrospective study design, a selection bias or other confounding factors cannot be excluded. Third, the histologic subtype of DLBCL was not considered because of the small size of the entire cohort hampering an otherwise statistical analysis. Our observations have to be confirmed by larger prospective studies using multivariate analyses which include tumor volumetric and serologic markers.

In conclusion, volume-based PET parameters derived from PET/CT and CT-textural features have the potential to predict therapy response in patients with DLBCL undergoing CAR-T cell therapy.

**Author Contributions:** Conceptualization, C.P.R. and M.H.; Data curation, C.P.R.; Formal analysis, C.P.R.; Investigation, M.H.; Methodology, C.P.R., C.F., H.D. and W.A.B.; Project administration, M.H.; Resources, C.L., K.N. and W.A.B.; Supervision, C.L., K.N., W.A.B. and M.H.; Validation, R.M.P. and H.D.; Writing—original draft, C.P.R. and M.H.; Writing—review & editing, R.M.P., C.F., C.L., K.N., H.D. and W.A.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Our study protocol was approved for retrospective evaluation of patient data by our institutional ethics committee with a waiver of the informed consent requirement (Project number 277/2020BO2).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.

**Conflicts of Interest:** Marius Horger received institutional research funds and speaker's honorarium from Siemens Healthineers and is a scientific advisor of Siemens Healthcare Germany. The other authors have declared that no competing interests exist.
