Treatment Assessment of pNET and NELM after Everolimus by Quantitative MRI Parameters

Assessment of treatment response to targeted therapies such as everolimus is difficult, especially in slow-growing tumors such as NETs. In this retrospective study, 17 patients with pancreatic neuroendocrine tumors (pNETs) and hepatic metastases (NELMs) (42 target lesions) who received everolimus were analyzed. Intralesional signal intensities (SI) of non-contrast T1w, T2w and DCE imaging, and apparent diffusion coefficients (ADCmean and ADCmin) of DWI, were measured on baseline and first follow-up MRI after everolimus initiation. Response assessment was categorized according to progression-free survival (PFS), with responders (R) showing a PFS of ≥11 months. ADCmin of NELMs decreased in Rs whereas it increased in non-responders (NR). Percentual changes of ADCmin and ADCmean differed significantly between response groups (p < 0.03). By contrast, ADC of the pNETs tended to increase in Rs, while there was no change in NRs. Tumor-to-liver (T/L) ratio of T1 SI of NELMs increased in Rs and decreased in NRs, and percentual changes differed significantly between response groups (p < 0.02). T1 SI of the pNETs tended to decrease in Rs and increase in Ns. The quotient of pretherapeutic and posttherapeutic ADCmin values (DADCmin) and length of everolimus treatment showed significant association with PFS in univariable Cox analysis. In conclusion, quantitative MRI, especially DWI, seems to allow treatment assessment of pNETs with NELMs under everolimus. Interestingly, the responding NELMs showed decreasing ADC values, and there might be an opposite effect on ADC and T1 SI between NELMs and pNETs.


Introduction
Neuroendocrine tumors of the pancreas (pNETs) represent a very rare tumor entity, with an incidence of 0.48 per 100.000 population and account for only 1% of all pancreatic tumors [1][2][3]. Due to the slow growth and the frequently unspecific symptoms-especially in the case of non-functional-pNETs-tumor stage is often advanced at the time of diagnosis, with hepatic metastases in around 64-70% of cases [4] Thus, curative tumor resection is limited, and therapy concepts are based on symptom control and reduction of tumor spread [5].
One of the most promising new therapy strategies for advanced NETs is based on the targeting and inhibition of the mTOR protein [6]. The mTOR pathway plays an important role in the tumorigenesis of NETs of different origins [1,6]. Due to its antiproliferative and antiangiogenic effects, everolimus, an oral inhibitor of the mTOR pathway, nowadays represents an integral option in the second-line therapy for progressive disease under SSAs or cytotoxic chemotherapy [7,8].
The RADIANT-3 trial proved the efficacy of everolimus in advanced pancreatic NETs, prolonging median PFS (11 months vs. 4.6 months for placebo group) and improving overall survival (OS) by 6.3 months (44 vs. 37 months) [1,9,10]. However, objective tumor response rates were generally low (5% in the everolimus group), indicating that the positive effect on PFS was mainly due to stabilization of tumor growth or minor tumor shrinkage, not reaching the cut-off for partial response as defined by RECIST [9,11]. An increasing body of evidence suggests that conventional response criteria, which are mainly based on tumor size change, are suboptimal to evaluate treatment response to antiproliferative or antiangiogenic effects as mediated by targeted anti-cancer drugs, especially in slowgrowing tumors such as NETs [11][12][13]. A recently published study showed that the Choi criteria, which integrate changes in tumor density in addition to size, correlated better with OS than RECIST in the therapy evaluation of pNETs under sunitinib [14]. Functional imaging parameters, especially diffusion-weighted imaging (DWI), provide encouraging capabilities to serve as new imaging biomarkers [15][16][17][18][19][20][21]. The European Neuroendocrine Tumor Society (ENETS) has already discussed ADC as a promising (despite little evidence so far) predictive parameter for tumor response in its expert consensus guideline and recommended the more frequent use of ADC in addition to RECIST [22]. Studies evaluating morphological response criteria of patients with NETs receiving everolimus or sunitinib are largely missing. However, this would be of fundamental importance, to allow an early differentiation of responders from non-responders in clinical practice and improve oncological study design by a more accurate definition of PFS.
Therefore, the purpose of this study was to analyze the diagnostic reliability of quantitative MRI for therapy response assessment in pNETs under everolimus.

Patients
Consecutive patients with histologically proven pNETs (resected or advanced) who received everolimus at our department, with therapy start between December 2011 and August 2021, were retrospectively included in this study. Further inclusion criteria were primary or secondary liver metastases and a baseline and follow-up MRI with DWI (Figure 1). Patient selection for therapy with everolimus was based on a consensus decision in an interdisciplinary tumor conference certified for NETs (ENETS Center of Excellence). The local research ethics committee (LMU Munich, project number 20-0847) approved this retrospective study and the need for written informed patient consent was waived.

MR Imaging
MR examinations were obtained using a 1.5T and 3T scanner (Magnetom Avanto, Magnetom Aera, Magnetom Skyra and Magnetom Verio, Siemens Healthcare, Erlangen, Germany). The imaging protocol consisted of an unenhanced T1w gradient-echo (GRE) sequence in-and out-of-phase, a single shot T2w sequence T1w 3D GRE sequences with fat suppression (fs) (non-contrast and 20, 50, 120 s after intravenous contrast media injection (Gd-EOB-DTPA; Primovist, Eovist, Bayer Schering Pharma, Germany; 25 µmol/kg body weight) depending on circulation time), a multishot T2w turbo spin echo sequence with fat saturation, diffusion-weighted sequences (b-values: 50, 400, 800 s/mm 2 ) and a T1w GRE sequence with fat saturation and a T1w 3D GRE fs sequence after a delay of 15 min. ADC maps were generated from DW images, including all b-values.

MR Imaging
MR examinations were obtained using a 1.5T and 3T scanner (Magnetom Avan Magnetom Aera, Magnetom Skyra and Magnetom Verio, Siemens Healthcare, Erlang Germany). The imaging protocol consisted of an unenhanced T1w gradient-echo (G sequence in-and out-of-phase, a single shot T2w sequence T1w 3D GRE sequences w fat suppression (fs) (non-contrast and 20, 50, 120 s after intravenous contrast media in tion (Gd-EOB-DTPA; Primovist, Eovist, Bayer Schering Pharma, Germany; 25 µmol body weight) depending on circulation time), a multishot T2w turbo spin echo seque with fat saturation, diffusion-weighted sequences (b-values: 50, 400, 800 s/mm 2 ) and a T GRE sequence with fat saturation and a T1w 3D GRE fs sequence after a delay of 15 m ADC maps were generated from DW images, including all b-values.

Image Analysis
All pretherapeutic MRI data were reviewed by two radiologists in consensus (C and S.K., with 15 and 2 years of experience in abdominal MRI). In consensus, they defin two hepatic metastases (minimum size > 0.5 cm) per patient as target lesions and asses if the pNET was resected. Inclusion criteria for metastases were homogenous appeara and no artifacts within the lesion in all sequences. Quantitative MRI evaluation was c ducted in two separate sessions by two readers independently: (1) preinterventional M and (2) postinterventional MRI with a 2-week interval between the review sessions. T readers were blinded for the patients' clinical or follow-up data. ADC values were cal lated by manually placing a circular region-of-interest (ROI) on the slice with the larg tumor extent on DWI (excluding structures close to the rim to avoid partial volume effe and on non-enhanced T2w and T1w + contrast-enhanced images (arterial, portal-veno venous, and hepatobiliary phase), SI values were recorded by drawing ROIs of the lesi as large as possible. In addition, ADC, T2w and T1w SI values of normal liver, pancr and spleen were measured on pre-and posttherapeutic scans, by placing circular ROIs large as possible, in areas of tumor-free tissue. The size (longest diameter) of hepatic

Image Analysis
All pretherapeutic MRI data were reviewed by two radiologists in consensus (C.S. and S.K., with 15 and 2 years of experience in abdominal MRI). In consensus, they defined two hepatic metastases (minimum size > 0.5 cm) per patient as target lesions and assessed if the pNET was resected. Inclusion criteria for metastases were homogenous appearance and no artifacts within the lesion in all sequences. Quantitative MRI evaluation was conducted in two separate sessions by two readers independently: (1) preinterventional MRI and (2) postinterventional MRI with a 2-week interval between the review sessions. The readers were blinded for the patients' clinical or follow-up data. ADC values were calculated by manually placing a circular region-of-interest (ROI) on the slice with the largest tumor extent on DWI (excluding structures close to the rim to avoid partial volume effects) and on non-enhanced T2w and T1w + contrast-enhanced images (arterial, portal-venous, venous, and hepatobiliary phase), SI values were recorded by drawing ROIs of the lesions as large as possible. In addition, ADC, T2w and T1w SI values of normal liver, pancreas and spleen were measured on pre-and posttherapeutic scans, by placing circular ROIs, as large as possible, in areas of tumor-free tissue. The size (longest diameter) of hepatic lesions was measured on T1-weighted hepatobiliary phase and on T1w arterial phase for the pNETs.

Standard of Reference and Response to Treatment
For all patients in the study cohort, diagnosis of pNET was confirmed by histopathology, and for most patients Ki-67 labelling index of the primary tumor was obtained. Grading was performed according to the 2017 WHO Tumor Classification Guideline (G1: Ki-67 Index was <3%, G2: Ki-67 Index was 3-20%, and G3 NET/NEC: Ki-67 Index was >20%) [23]. Treatment response was evaluated as progression-free survival (PFS) and was calculated in months from the time of everolimus initiation until progression according to the local interdisciplinary tumor board, including evaluation of all performed imaging studies (CT, PET/CT, MRI). Responders were defined by PFS ≥ 11 months and non-responders (NRs) by PFS < 11 months, respectively. This cut-off was based on the reported median PFS of 11 months in the RADIANT-3 phase-III-trial. Additionally, OS was determined in months Biomedicines 2022, 10, 2618 4 of 13 from treatment start until death from any cause. Patients who were still alive at the time of last follow-up (May 2022) were censored.

Statistical Analysis
Statistical analysis was performed utilizing commercially available statistical software (Graphpad Prism Version 6, San Diego, Calif. and SPSS version 25, Chicago, IL, USA). Statistical significance level was defined at p ≤ 0.05. All data was expressed as mean or median values with standard deviation (SD) or interquartile range [IQR], respectively. Normal distribution of continuous variables was evaluated by visual inspection of the frequency distribution (histogram). We compared the SI and ADC values of target lesions before and after therapy using a two-tailed, paired t-test. Two-tailed, unpaired t-tests were used to compare the SI and ADC values of target lesions between different response groups, respectively. The Wilcoxon rank-sum test was used in case of a non-normal distribution. OS and PFS were modeled according to the Kaplan-Meier method. The multivariable Cox proportional hazards regression model was performed to analyze independent prognostic clinical and imaging parameters for PFS and OS. Using a stepwise approach, all variables with p < 0.1 in univariable analysis were entered into the multivariable model.

Patients
Seventeen patients (6 female, 11 male) with a total of 42 target lesions (34 liver metastases, 8 primary tumors) were included in this retrospective study. Due to motion artifacts, DWI was not evaluated in one patient and two pNETs were not evaluated due to intratumoral calcifications and heterogenous appearance, respectively. Baseline MRI was acquired 26 ± 32 d prior to treatment start, and first follow-up MRI was performed 115 ± 44 d after initiation of everolimus. A total of 7/17 patients underwent primary tumor resection. Most of the included NET patients had G2 tumors (n = 14), followed by two high-grade tumors (G3), while no low-grade (G1) tumors were included, and for one patient no grading was obtained. Detailed patient characteristics are presented in Table 1.

ADC Measurements
There were no significant differences of ADC values of NELMs between Rs and NRs before treatment. ADCmin in responding NELMs decreased significantly after therapy (p < 0.02), while ADCmin increased in non-responding NELMs (p< 0.03) (Figure 2A     interventional axial T1w image (hepatobiliary phase) shows hypointense metastasis (circle) in segment 2. The metastasis (arrow) shows restricted diffusion with high signal on axial DW-MR image b = 800 s/mm 2 (B,D) dark signal (circle) on ADC map (C). The pretherapeutic ADCmin value of the metastasis was 858 × 10 -3 mm 2 /s. Under everolimus, the metastasis showed no significant change in size on hepatobiliary phase (D), however there was a loss of signal on DW-MR image b = 800 s/mm 2 (E) compared to the pre-interventional image and decreasing signal (circle) on the ADC map (F). The post-interventional ADCmin values were 476 × 10 -3 mm 2 /s and lower higher compared to pretherapeutic ADC value. Regarding the primary tumor, statistical analysis was limited due to the small sample size. In Rs, both ADCmin and ADCmean tended to increase after therapy initiation, while there was no change in NRs ( Figure 2C). For ADCmean, percentual change was 37.1% (IQR 25.6-48.5) in Rs compared to 1.2% (IQR −13.4-25) in NRs.
There were no significant changes between the pre-and posttherapeutic ADCmean of non-tumorous liver, spleen, and pancreatic tissue.

T2 Signal
T2w of NELM divided by the T2w background signal of the spleen (T/S ratio) increased significantly in NRs while there was no change in Rs ( Figure 5). Regarding the primary tumor, statistical analysis was limited due to the small sample size. In Rs, both ADCmin and ADCmean tended to increase after therapy initiation, while there was no change in NRs ( Figure 2C). For ADCmean, percentual change was 37.1% (IQR 25.6-48.5) in Rs compared to 1.2% (IQR −13.4-25) in NRs.
There were no significant changes between the pre-and posttherapeutic ADCmean of non-tumorous liver, spleen, and pancreatic tissue.

Non-Enhanced T1 Signal
The ratio of T1 SI of the target lesion divided by T1 SI of the liver (T/L ratio) showed a significant increase in the R group after therapy start, while T/L ratio tended to decrease in the NR group ( Figure 6A). Concordantly, percentual changes of non-enhanced T/L ratio differed significantly between response groups (p = 0.02) with an increase of 38.6% in the R group (IQR 31.8-76.1), compared to a decrease of −24.2% in the NR group (IQR −50-9.9) ( Figure 6B). Non-enhanced T1 of the pNET tended to decrease in Rs and slightly increase in NRs; however, statistical analysis was not possible due to the small number of included pNETs. Native TI SI of non-tumorous liver and pancreas did not differ significantly between groups, and after the start of everolimus.

Non-Enhanced T1 Signal
The ratio of T1 SI of the target lesion divided by T1 SI of the liver (T/L ratio) showed a significant increase in the R group after therapy start, while T/L ratio tended to decrease in the NR group ( Figure 6A). Concordantly, percentual changes of non-enhanced T/L ratio differed significantly between response groups (p = 0.02) with an increase of 38.6% in the R group (IQR 31.8-76.1), compared to a decrease of −24.2% in the NR group (IQR −50-9.9) ( Figure 6B). Non-enhanced T1 of the pNET tended to decrease in Rs and slightly increase in NRs; however, statistical analysis was not possible due to the small number of included pNETs. Native TI SI of non-tumorous liver and pancreas did not differ significantly between groups, and after the start of everolimus.
The ratio of T1 SI of the target lesion divided by T1 SI of the liver (T/L ratio) showed a significant increase in the R group after therapy start, while T/L ratio tended to decrease in the NR group ( Figure 6A). Concordantly, percentual changes of non-enhanced T/L ratio differed significantly between response groups (p = 0.02) with an increase of 38.6% in the R group (IQR 31.8-76.1), compared to a decrease of −24.2% in the NR group (IQR −50-9.9) ( Figure 6B). Non-enhanced T1 of the pNET tended to decrease in Rs and slightly increase in NRs; however, statistical analysis was not possible due to the small number of included pNETs. Native TI SI of non-tumorous liver and pancreas did not differ significantly between groups, and after the start of everolimus.

DCE
There were no significant differences in the contrast enhancement of NELM between arterial, pv and venous phases between response groups. However, in accordance with the results of non-enhanced T1w SI evaluation, background T1 T/L ratio after therapy start tended to have higher SI ratios in responding lesions than non-responding NELMs. For arterial phase, T/L ratio was 1.18 in Rs compared to 0.86 in NRs (p < 0.05) (Figure 7).

DCE
There were no significant differences in the contrast enhancement of NELM between arterial, pv and venous phases between response groups. However, in accordance with the results of non-enhanced T1w SI evaluation, background T1 T/L ratio after therapy start tended to have higher SI ratios in responding lesions than non-responding NELMs. For arterial phase, T/L ratio was 1.18 in Rs compared to 0.86 in NRs (p < 0.05) (Figure 7).

Cox Regression of PFS and OS
In univariable analysis of PFS, length of everolimus treatment and the imaging parameter DADCmin, which represents the quotient of ADCmin of NELM before and after therapy start (ADCmin pretherapeutic/ADCmin posttherapeutic = DADCmin), showed a significant association with PFS. The other clinical parameters, including age, ki-67, grading and elevated baseline NSE or Chromogranin A levels showed no significant associations with PFS in the univariable analysis (Table 2). In the multivariable model, neither of both parameters remained significant; however, the p-value of DADCmin was borderline (0.09), especially regarding the small cohort for Cox analysis. In the univariable analysis for OS, none of the parameters showed significant association with patient outcome (Table  3).

Cox Regression of PFS and OS
In univariable analysis of PFS, length of everolimus treatment and the imaging parameter DADCmin, which represents the quotient of ADCmin of NELM before and after therapy start (ADCmin pretherapeutic/ADCmin posttherapeutic = DADCmin), showed a significant association with PFS. The other clinical parameters, including age, ki-67, grading and elevated baseline NSE or Chromogranin A levels showed no significant associations with PFS in the univariable analysis (Table 2). In the multivariable model, neither of both parameters remained significant; however, the p-value of DADCmin was borderline (0.09), especially regarding the small cohort for Cox analysis. In the univariable analysis for OS, none of the parameters showed significant association with patient outcome (Table 3).

Discussion
In this study, we found that quantitative MRI parameters, especially ADC and T1w SI of NELMs, were useful for response assessment of pNETs treated by the targeted agent everolismus. We chose PFS as the primary end point, as recommended by the National Cancer Institute Neuroendocrine Tumor Clinical Trials Planning Meeting consensus report. The use of OS is especially challenging in NETs due to generally long survival times, and consequently a wider range of post-progression salvage therapies [4]. Overall median PFS was rather low in our study cohort, with 5 months compared to a median PFS of 11 months in the RADIANT-3 trial [9]. Overall median survival was also slightly lower than reported, with 36 months in our study compared to 44 months in the RADIANT-3 trial [4]. Factors which might cause this difference in our small study cohort in comparison to the RADIANT-3 trial data might be tumor grading, percentage of patients with high baseline NSE 1 × ULN or high baseline CgA 2 × ULN, as well as late treatment line with everolimus. In the RADIANT-3 trial, 82% had well-differentiated tumors and only 17% moderately differentiated tumors, while in our cohort 82% had moderately differentiated tumors and 12% low-differentiated tumors. In the RADIANT-3 trial chromogranin, A was elevated in 46% and NSE was elevated in 26% of all patients, while in our patient group CgA was elevated in 76% and NSE was elevated in 53% [24][25][26].
Interestingly, we found that ADCmin values showed a significant decrease after everolimus initiation in responding NELMs, while ADC in NRs increased. The quotient of ADCmin pretherapeutic/ADCmin posttherapeutic was the only imaging parameter that showed a significant association with PFS in univariable Cox analysis (HR 0.21 (95% CI 0.06-0.8; p = 0.02). This underlines that lower ADCmin values in follow-up MRIs were associated with better treatment response. This finding seems counterintuitive at first sight, as therapies inducing apoptosis usually show an increase in ADC values due to tumor necrosis and lysis; however, it depends on the mechanism of treatment. For example, PRRT in neuroendocrine tumors has been reported to cause increased ADC values in patients with better OS and PFS [16], and the percentual increase of ADC was found to correlate with decrease in size [18].
The effect of targeted, cytostatic agents is not immediate cell death as in cytotoxic regimens. pNETs characteristically contain a high microvascular density, which is expressed by radiological enhancement [14]. Previous studies investigating targeted therapies with antiangiogenic effects have shown (a transient) decrease in tumor ADC values, which might be accompanied by reduced contrast enhancement.
It is likely that antiangiogenic effects reduce vascular permeability, which is accompanied by lower perfusion and extravascular-extracellular space, and thus decreasing ADC. Only larger tumor necrosis mediated by vascular-directed therapies would lead to an increase of ADC [11,27].
Schraml et al. investigated early changes of DWI in advanced hepatocellular carcinoma treated with the multikinase inhibitor sorafenib and found an early decrease in ADC, followed by a later re-increase. The authors suggested that these changes were caused by hemorrhagic tumor necrosis. Antiangiogenic treatments induce ischemic conditions, which causes a disruption of the cell membrane and in turn leads to a shift of extracellular ions and water molecules into the cell while reducing the extracellular volume [11,28,29].
Additionally, for patients who received anti-VEGF drugs in brain tumors, a transient decrease of ADC was reported [30,31]. Patients with renal cell carcinoma, which is also a highly vascular tumor, treated with sunitinib showed an early reduction of D (IVIM) of the primary tumor, with a larger reduction being associated with a better RECIST response and longer PFS [32]. It was shown that in spite of increased histological necrosis, the radiological-assessed diffusion restriction increased [33].
Another influencing factor could be related to the immunological effects, as everolimus is known to increase the rate of regulatory T-cells [34]. Additionally, increased fibrosis due to replacement of viable tumor cells by collagen was hypothesized as a cause for decreasing ADC values in responding hepatic metastases indicated by improved OS [35].
For evaluation of T1 SI acquisition parameters are not standardized and studies evaluating comparability are missing. Native T1 SI of NELM decreased in NR and increased in R-both for absolute values as well as for the relative ratio of T1 SI of the target lesion divided by T1 SI of the liver (T/L ratio). NELM in the R group showed an increase of around 39% after everolimus initiation (vs. −24% for NR). Two different studies, both analyzing HCC, found hyperintense SI of non-enhanced T1w after treatment with sorafenib [28,36], suggesting that therapy induced intralesional hemorrhage or protein-rich necrosis. Mostly, no intratumoral (macro-) hemorrhage was detected in T1w images in our study cohort, but this might be explained by microbleeding, which is not visually noticeable but affects T1w SI and DWI, and according to the time-dependent evolution of blood products, the hemoglobin might be obscured by the fat saturation [28].
Elevated baseline levels of neuron-specific enolase (NSE) and chromogranin A (CgA) were shown to be prognostic for poor survival in pNETs under everolimus [1,37]. In our study cohort, DADCmin correlated better with OS than baseline CgA, while there was no significant correlation with pretherapeutic NSE levels, which might be explained by a small sample size.
Tumor angiogenesis is supposed to play a crucial role in response assessment of targeted therapies [12]. Regarding the quantitative DCE measurements, we found no significant changes in tumor enhancement. However, the analysis of enhancement is prone to errors as it depends on the exact timing of contrast-enhanced images, and different models for tracer kinetics are used by different research groups. In this study, DCE (including four contrast phases) was used instead of a high-resolution over-time perfusion technique. In addition, quantitative DCE evaluation, e.g., quantitative EASL (European Association for the Study of Liver) is still labor-and time-intense, and thus not suitable for clinical routine [38].
Evaluation of ADC and non-enhanced SI of T1w of the pNETs tended to show opposite effects compared to the NELM. T1w SI of the pNET tended to decrease in R and slightly increase in NR, and ADC (min and mean) of the pNETs tended to increase in R, while there was no change in NR. However, statistical analysis was not possible due to the small number of included pNETs, as they were either resected (n = 7) or strong intratumoral heterogeneities impeded quantitative evaluation. Thus, results of the quantitative evaluation of pNETs should not be overinterpreted; however, we included the data for possible future systematic reviews.
One major limitation of this study is the small sample size, especially of pNETs, due to the low incidence of pNETs and everolimus representing a second-line treatment. As the study design was retrospective, time intervals were not entirely homogenous and imaging data were not acquired on a single MRI scanner. Therefore, for T1w and T2w SI measurement, the ratio of the background liver/spleen was calculated to eliminate scannerdependent confounding factors. DWI represents simultaneous information regarding diffusion and microperfusion, so it can be difficult to attribute the observed changes to one specific pathomechanism [29]. The use of IVIM, which offers quantitative perfusion and diffusion evaluation, could potentially be helpful for future studies.

Conclusions
Treatment response of targeted therapies such as everolimus might be better depicted by changes in quantitative MRI parameters than conventional size-based criteria. As suggested by the low objective response rates achieved in the RADIANT trials, everolimus only mediates minor tumor shrinkage [4,9]. Our results suggest that the ADC and SI of T1w represent valuable parameters to monitor changes after everolimus treatment and might be able to better differentiate responding from non-responding patients. Interestingly, NELMs of the R group showed a decrease of ADC, which is opposite to the reported changes of DWI after chemotherapy and radiation. However, smaller studies analyzing hypervascular tumors (HCC, RCC) treated by targeted agents also reported decreasing ADC values after therapy. Regarding the major role of imaging response assessment in the development of new treatment strategies, these observations should be further investigated in prospective studies, in a larger multi-center setting.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of LMU Munich (project number 20-0847).
Informed Consent Statement: Patient consent was waived due to retrospective study design.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.