Longitudinal monitoring of Apparent Diffusion Coefficient (ADC) in patients with prostate cancer undergoing MR-guided radiotherapy on an MR-Linac at 1.5 T: a prospective feasibility study

All patients included in this prospective study were recruited in the M-base Pro 1.0⁹ or M-base HyPro 2.0 at our institution (ClinicalTrials.gov Identifiers: NCT02724670; NCT03880851). The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of the medical faculty of Tuebingen University (No. 022/2016BO1, 14.03.2016 and No. 920/2018BO1, 10.07.2019). Informed consent was obtained from all subjects involved in the study. All patients consented to prospectively undergo multiple MRIs on an MRL and additionally on a MRI_3T at several points prior to and during RT. The aforementioned studies each examine a novel MR-adaptive concept for radiotherapy of primary localized prostate cancer. Between February 2019 and October 2021, 9 patients with biopsy-confirmed prostate cancer and available MRL and MRI_3T data sets prior to RT and under RT were included. All patients were treated daily on a 1.5T MRL (Elekta Unity^TM, Philips, Stockholm, Sweden).¹⁰ All patients were treated according to national guidelines with either 39 × 2 Gy per fraction over eight weeks (M-base 1.0 study, n = 3 patients) or 20 × 3 Gy per fraction over four weeks (M-base Hypro 2.0 study, n = 6 patients) and additional neoadjuvant androgen deprivation therapy (ADT) of six months for intermediate risk patients and 24–36 months for high risk patients.¹¹ The time point of the MRL and MRI_3T was during week 2 of RT in both treatment protocols. MRI technique, specifications and acquisition parameters of the examinations on both systems have been previously described.⁸ Most study participants in this study were used in the prior publication⁸, but only examinations prior to RT were analyzed. No lesion ADC dynamics during RT were reported in the previous study.⁸

The ADC maps for MRL and MRI_3T for each patient, prior to and during radiotherapy, were independently presented to two readers (reader 1, a board certified radiation oncologist with 8 years of experience reader 2, a radiology resident with 4 years of experience). Both readers placed an elliptic region-of-interest (ROI) within the lesions for each patient in MRL and MRI_3T image sets. A dedicated workstation (GE Healthcare Centricity™ PACS RA1000, Milwaukee WI, USA) was utilized for image analysis using a dedicated software (syngo.via, Siemens Healthcare, Erlangen, Germany).

Continuous variables were reported as mean and standard deviation. Paired t-tests were used for pair-wise pre- vs. during-treatment comparisons, as well as MRL vs. MRI-3T. Intraclass correlation coefficient (ICC, two-way, absolute agreement) was used to compute inter-reader agreement. An ICC of less than 0.4 signalizes poor agreement, of 0.40 to 0.59 indicates fair agreement, of 0.60 to 0.74 good agreement, and an ICC of 0.75 to 1.00 signalizes excellent agreement.¹² Pearson Correlation coefficient was used to compare lesion ADC between MRL and MRI_3T. Level of significance was set at 0.05. Statistical analyses were performed using SPSS (v26.0, IBM-Corp, Armonk, NY, USA).

ADC changes of the intraprostatic tumor are to be expected both during radiotherapy and during ADT and the correlation between ADC and prostate cancer aggressiveness has been shown before on diagnostic scanners.^17,18 The mean ADC-values calculated in this study on both scanners are similar to values found in the literature: Tamada et al. reported mean ADC values of (untreated) tumor regions of 1.02 ± 0.25 × 10⁻³ mm²/s for the peripheral zone and of 0.94 ± 0.21 × 10⁻³ mm²/s for the transitional zone of the prostate.¹³ Van Schie et al. reported median ADC values in the tumor scanned on a diagnostic scanner of 1.08 ± 0.39 × 10⁻³ mm²/s (mean ± SD) prior to treatment and assessed changes of ADC prior to a hypofractionated RT and then weekly during RT in 73 patients in a similar manner as performed in our study. The group found a (non-significant) median increase of the ADC-value in the tumor of 7% for patients with concurrent ADT and a median increase of 20% for patients without ADT.¹⁴

Moreover, ADC values and -changes have been shown to function as biomarkers for RT response in prostate cancer patients.^6,15 Radiomics approaches seem promising in assessing response to RT, as performed by Abdollahi et al. prior to vs. after RT.¹⁶ In all of these studies and in prostate cancer diagnostics, a 3 T MRI scanner has been established as the gold standard for mpMRI.¹⁷ MRI_3T leads to optimal diagnostic images, often aided by suppression of peristalsis via intravenous application of butyl scopolamine or other agents. In contrast, on a 1.5 T MRL, the utilized sequences are optimized for fast and geometrically accurate image acquisition in an online workflow without routine administration of peristalisis suppressing medications or contrast agents. These possible limitations, in addition to technical differences of the hybrid system to diagnostic scanners¹⁸, pose the question whether an MRL can deliver comparable functional information during RT of prostate cancer.

In principle, the hybrid system offers fertile ground for further plan adaption in prostate carcinoma patients based on mpMRT findings such as ADC values since it offers daily MR-guided plan adaptations. Treatment individualization and plan adaptation under RT are a focus of research in other tumor entities as well.^7,19,20 Longitudinal diffusion MRI on a 0.35 T hybrid system was already performed in small series for several other tumor entities with promising results (Yang et al.: three head and neck cancer patients and three sarcoma patients²¹; Shaverdian et al.: three rectal cancer patients²²). On the 1.5 T MRL, Lawrence et al. report a high ADC repeatability and comparability to a diagnostic 1.5 T scanner for 59 patients with central nervous system tumors.²³ Habrich et al. used a test-retest approach on 11 patients with head and cancers and showed to a high repeatability of ADC measurements on the 1.5 T MRL.²⁴ In the context of prostate carcinoma, Habrich et al. also examined intravoxel incoherent motion (IVIM) and dynamic contrast enhanced (DCE) MRI changes over the course of a moderately hypofractionated RT in 20 patients, also indicating that longitudinal measurements of functional imaging parameters is feasible and could be used for response assessment in the future.²⁵

However, concerning intraprostatic tumor lesions, the verification of longitudinal stability of ADC measurements on an MRL as performed in this study in a comparison to a latest generation 3 T diagnostic scanner has to the best of our knowledge not been performed yet.

In a previous study, we tested the clinical applicability based on qualitative and quantitative parameters of prostate MR images on an MRL against a MRI_3T at one point of time prior to starting RT. We were able to show a promising and comparable result of T2 weighted image quality, and lesion conspicuity and we reported comparable lesion ADC measurements between MRL and MRI_3T.⁸

With the current work, we demonstrate longitudinal comparability and reliability between the two system during RT. This represents a necessary basis for future analyses of lesion changes over time on the MRL and confirms its potential for individualized treatment adaptations such as dose painting²⁶ and response assessment during treatment.

This study has limitations. Firstly, the small sample size of patients who underwent multiple prostate imaging at both devices and the fact that only one time-point during radiotherapy was used for analysis. Multiple time-points during the course of radiotherapy should be analyzed to further validate the stability and comparability of ADC measurements. However, logistic challenges hindered further validation with an MRI_3T at more than one time point. Secondly, treatment regimens differed in this population (either 20 × 3 Gy oder 39 × 2 Gy, additional neoadjuvant ADT in 3 patients). Thirdly, the DWI acquisition parameters did not fully conform to the published recommendations of the MR-linac consortium, which were published after we had already included the patients in our study and predefinded the technical aspects of the utilized sequences.²⁷

Nonetheless, this study depticts the reality and the challenges of clinical routine and its preliminary findings could be considered novel. Indeed, further prospective studies examining mpMRI data under RT and correlating those with clinical endpoints are desirable to advance individualized radiation treatment.

In conclusion, lesion ADC as measured on MRL increased significantly during radiotherapy and lesion ADC measurements on both systems showed similar dynamics. These preliminary findings are promising but need large-scale validation. Once validated, lesion ADC on MRL might be used as a biomarker for real-time assessment of tumor response in patients with prostate cancer undergoing MR-guided radiation therapy.