Prognostic Value of Preoperative Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Epithelial Ovarian Cancer

Kokoteksti

(1)UEF//eRepository DSpace Rinnakkaistallenteet. https://erepo.uef.fi Terveystieteiden tiedekunta. 2019. Prognostic Value of Preoperative Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Epithelial Ovarian Cancer Lindgren, Auni Elsevier BV Tieteelliset aikakauslehtiartikkelit © Elsevier B.V. CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ http://dx.doi.org/10.1016/j.ejrad.2019.03.023 https://erepo.uef.fi/handle/123456789/7629 Downloaded from University of Eastern Finland's eRepository.

(2) Accepted Manuscript Title: Prognostic Value of Preoperative Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Epithelial Ovarian Cancer Authors: Auni Lindgren, Maarit Anttila, Otso Arponen, Suvi Rautiainen, Mervi Könönen, Ritva Vanninen, Hanna Sallinen PII: DOI: Reference:. S0720-048X(19)30125-1 https://doi.org/10.1016/j.ejrad.2019.03.023 EURR 8518. To appear in:. European Journal of Radiology. Received date: Revised date: Accepted date:. 4 January 2019 20 March 2019 29 March 2019. Please cite this article as: Lindgren A, Anttila M, Arponen O, Rautiainen S, Könönen M, Vanninen R, Sallinen H, Prognostic Value of Preoperative Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Epithelial Ovarian Cancer, European Journal of Radiology (2019), https://doi.org/10.1016/j.ejrad.2019.03.023 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain..

(3) Prognostic Value of Preoperative Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Epithelial Ovarian Cancer. TE D. M. A. N. U. SC R. IP T. Auni Lindgren MDa a University of Eastern Finland, Faculty of Health Sciences, School of Medicine, Institute of Clinical Medicine, Obstetrics and Gynecology Maarit Anttila MD PhDa,b a University of Eastern Finland, Faculty of Health Sciences, School of Medicine, Institute of Clinical Medicine, Obstetrics and Gynecology b Department of Gynecology, Kuopio University Hospital, Kuopio, Finland maarit.anttila@kuh.fi Otso Arponen MDc c Department of Clinical Radiology, Kuopio University Hospital, Kuopio, Finland otso.arponen@kuh.fi Suvi Rautiainen MD PhDc c Department of Clinical Radiology, Kuopio University Hospital, Kuopio, Finland suvi.rautiainen@icloud.com Mervi Könönen FM PhDc,d c Department of Clinical Radiology, Kuopio University Hospital, Kuopio, Finland d Department of Clinical Neurophysiology, Kuopio University Hospital, Kuopio, Finland mervi.kononen@kuh.fi Ritva Vanninen MD PhDc,e,f c Department of Clinical Radiology, Kuopio University Hospital, Kuopio, Finland e University of Eastern Finland, Faculty of Health Sciences, School of Medicine, Institute of Clinical Medicine, Clinical Radiology f University of Eastern Finland, Cancer Center of Eastern Finland, University of Eastern Finland Kuopio, Finland ritva.vanninen@kuh.fi Hanna Sallinen MD PhDb b Department of Gynecology, Kuopio University Hospital, Kuopio, Finland hanna.sallinen@kuh.fi. EP. Work originated: Kuopio University Hospital, P.O. Box 100, Kuopio. A. CC. Corresponding author: Auni Lindgren, M.D. Department of Gynecology and Obstetrics Kuopio University Hospital P.O. Box 100, Kuopio, FIN-70029 KYS, Finland tel. +358 45 1481008, Fax +358 17 172 383, Email: Auni.lehikoinen@gmail.com. Highlights . DCE-MRI parameters may provide preoperative imaging biomarkers in ovarian cancer. 1.

(4) . Mean Ktrans values were higher in high-grade serous OC than in other types. . Lower mean Ktrans values predicted residual tumor in sytoreductive surgery. . Multiparametric MRI variables were predictive for recurrence free survival. IP T. Abstract Objectives. SC R. To investigate whether semi-quantitative and pharmacokinetic perfusion dynamic contrast-enhanced (DCE) parameters are associated with traditional prognostic factors and can predict clinical outcome in ovarian cancer (OC).. U. Methods. N. This prospective study, approved by local ethical committee, enrolled 38 patients with primary OC, 2011– 2014. After preoperative DCE-MRI (3.0 T), two observers measured perfusion (Ktrans, Kep, Ve, Vp) and semi-. A. quantitative parameters (area under the curve, peak, time-to-peak) by drawing regions of interest (ROIs). M. covering the large solid lesion (L-ROI) and the most enhancing small area (S-ROI) (NordicICE platform).. TE D. Kruskal–Wallis was used to analyze associations between MRI parameters and classified prognostic factors. Results. Mean Ktrans values were higher in high-grade serous OC than in other types (L-ROI, P = 0.041; S-ROI, P =. EP. 0.018), and lower mean Ktrans values predicted residual tumor (L-ROI P = .030; S-ROI, P = 0.012). Higher minimum Vp values were associated with higher International Federation of Gynecology and Obstetrics. CC. (FIGO) stage (S-ROI, P = 0.023).Shorter recurrence-free survival was predicted by higher Ve (minimum LROI, P = 0.035; maximum S-ROI, P = 0.046), Vp (maximum S-ROI, P = 0.033), and lower time-to-peak. A. (maximum S-ROI, P = 0.047) in Kaplan–Meier analysis. Multiparametric MRI variables combining DCE and diffusion weighted data were also predictive for survival. Conclusion DCE-MRI parameters may represent imaging biomarkers for predicting tumor aggressiveness and prognosis in OC. Higher Ktrans levels were associated with better results in cytoreductive surgery but with earlier recurrence.. 2.

(5) Abbreviations EES = extravascular, extracellular space; Ktrans = a rate constant for transfer of contrast agent from plasma to EES ; Kep = rate constant for transfer of contrast agent from EES to plasma; Ve = contrast agent distribution volume, EES volume fraction; Vp = plasma volume fraction; AUC = area under the enhancement curve;. IP T. WashIn = initial up-slope of the DCE curve; WashOut = initial down-slope of the DCE curve; Peak = Peak / maximal enhancement; Time To Peak = time when contrast agent reaches the peak volume; AIF = arterial. SC R. input function, NACT = neoadjuvant chemotherapy, mpMRI = multiparametric MRI. Keywords. A. CC. EP. TE D. M. A. N. U. Ovarian Cancer; Magnetic resonance imaging; Dynamic Contrast-Enhanced Imaging; Biomarkers; Prognosis. 3.

(6) Introduction Ovarian cancer (OC) is the fifth most common cancer and fourth most common cause of cancer mortality in women[1]. Characterizing ovarian masses as precisely as possible is important because treatment modalities are tailored individually, and especially during a woman’s reproductive years, conservative surgery for fertility-sparing can be crucial. Vaginal ultrasonography is the initial modality for investigating ovarian. IP T. tumors, and the International Ovarian Tumor Analysis group guideline can be used to estimate the malignancy risks of ovarian tumors[2]. Approximately 20% of ovarian tumors remain indeterminate after an ultrasound conducted by a specialist. Also the risk of malignancy index (RMI) helps physicians differentiate. SC R. benign from malignant lesions[3]. Magnetic resonance imaging (MRI) is valuable especially in diagnostics of this indeterminate group in US and intermediate- and low-risk ovarian tumors (RMI < 200), giving better. U. soft tissue contrast than computed tomography (CT). Also in ovarian cancer diagnostics MRI can yield more robust information than CT. Diffusion weighted imaging (DWI) has shown promise in tumor staging,. N. predicting the aggressiveness of tumor and clinical outcome[4,5]. Cancer treatment is becoming. A. individualised, so it is important to further study the possibilities to obtain more information also in. M. diagnostic imaging. If it was possible to identify patients who can be operated optimally in sytoreductive. TE D. surgery and define the patient group with most aggressive tumors with preoperative imaging, it could be very valuable in the decision-making of treatment options. Dynamic contrast-enhanced (DCE) MRI is used to improve the diagnostic accuracy of conventional MRI,. EP. with proven importance in differential diagnostics and preoperative evaluation for breast, prostate, and kidney tumors, among others[6–8]. DCE-MRI can distinguish malignant from benign tumors based on. CC. differences in contrast agent behavior; in malignant tumors, the microcirculation is different because of neoangiogenesis [9,10]. Most DCE-MRI studies of ovarian tumors have targeted differentiating among. A. benign, borderline, and malignant tumors[11–14], and studies often have used semi-quantitative and time intensity curve–based parameters[11–13]. European Society of Urogenital Radiology guidelines advocate inclusion of DCE time intensity curve analysis to specify indeterminate ovarian masses[15]. Because the volume of transfer constant (Ktrans) level is influenced by blood flow and vessel permeability properties, it has been under active investigation to be used as a biomarker of tumor perfusion and permeability in DCE-MRI cancer studies[16–21], with contradictory results. Also other pharmacokinetic. 4.

(7) perfusion parameters reflect the physiology of circulation in the microvasculature and can be quantitatively compared among different patients and investigators[9,22]. Therefore, DCE imaging might provide more extensive prognostic information in OC, as well.. We hypothesized that DCE parameters will differ between highly aggressive OC and less severe disease. The. IP T. main objectives of the present study were to investigate whether semi-quantitative and perfusion DCE parameters are associated with traditional prognostic factors and able to predict the clinical course of OC.. Our secondary objectives were to investigate the possible inter-technique differences related to ROI size and. SC R. analyze interobserver variability in DCE measurements.. U. Materials and Methods Study protocol and patients. N. This exploratory prospective single-institution study was conducted between January 2011 and December. A. 2014. The local research ethical committee approved the study protocol, and written informed consent was. M. obtained. The inclusion criteria were a clinical diagnosis of primary OC, fallopian tube cancer, or peritoneal. TE D. carcinoma and measurable disease at staging CT. Exclusion criteria were contraindications to MRI or gadolinium contrast agents. A total of 38 patients were enrolled. These patients were used also in our previous article concentrating on ADC values[5]. In the current study we have analysed the possible. EP. predictive value of the DCE parameters, both alone and combined with ADC values, now with an extended follow-up in survival analysis. Cancers were staged using the International Federation of Gynecology and. CC. Obstetrics (FIGO) guidelines. Histological type and grade were evaluated according to World Health Organization criteria. First-line treatment was chosen by an experienced multidisciplinary team (surgery, n =. A. 34; neoadjuvant chemotherapy (NACT) before surgery, n = 4). Patients received paclitaxel–carboplatin as adjuvant chemotherapy after the operation, excluding a patient with stage 1A disease who received carboplatin monotherapy. Patient characteristics are described in Table 1.. Imaging protocol. 5.

(8) All patients underwent 3.0 T MR imaging (Philips Achieva 3.0T TX, Philips N.V., Eindhoven, The Netherlands) before treatments. The MRI protocol included T2w, T1w (non-contrast and contrast enhanced), diffusion-weighted imaging (DWI) and DCE sequences for dynamic analysis (Table 2). During DCE image acquisition, non-contrast images were scanned first (image stack 1), followed by contrast agent administration, and continued image acquisition (image stacks 2–23). The contrast agent gadoterate. IP T. meglumine (Dotarem® 279.3 mg/ml, Guerbet, France) was injected intravenously as a bolus dose of 0.1 mmol/kg at a rate of 4 ml/s, followed by a 20 ml flush of 0.9% sodium chloride solution using an MRI-. compatible power injector (Optistar Elite, Covidien, Los Angeles, CA, USA). T1w postcontrast images were. SC R. scanned 3D.. U. Image analysis. Structural MR images were evaluated independently and blinded to histological information by two. N. observers (**, **), with 4 and 3 years of experience in MRI, using a Sectra PACS workstation (IDS7,. A. Version 15.1.20.2, Sectra AB, Linköping, Sweden). A senior radiologist (**) with 11 years of experience. M. verified the identification and delineation of ADC measurements.. TE D. DCE-MRI parameter maps were generated automatically using NordicICE (version: 2.3.13, NordicNeuroLab, Bergen, Norway). Motion correction was done automatically. Five semi-quantitative parameter maps were generated from the DCE curve: area under the DCE curve; WashIn, the initial up-slope. EP. of the DCE curve; WashOut, down-slope of the DCE curve; peak, the amplitude of peak enhancement; and time-to-peak, the time when contrast agent reaches peak enhancement. To quantitate the perfusion the. CC. arterial input function was determined by using a small AIF ROI from the common or external iliac artery. Four quantitative parameter maps were generated: Ktrans, a rate constant for transfer of contrast agent from. A. plasma to the extravascular extracellular space (EES); Kep, the rate constant from EES to plasma; Ve, contrast agent distribution volume (% volume of EES per unit volume of tissue); Vp, plasma volume. Measurements were obtained using the transaxial image showing the largest solid tumor diameter in the ovary. In clinical practice two-dimensional measurement is practical and previous evidence from DW imaging supports its use[23].Two different regions of interest (ROIs) were defined: large ROI (L-ROI) was drawn free-hand to cover the whole solid tumor area, excluding necrotic, cystic, and vascular areas; small ROI (S-ROI), circle. 6.

(9) with a diameter of 6 to 12mm, was placed on the area considered most solid and high enhancement identified by visual assessment. T2w, T1w, DWI, and contrast-enhanced T1w sequences were all available for ROI localization and tumor demarcation (Figure 1). DWI helped to identify the most solid part while the early enhancement time point, the wash in phase, was used for assessment of enhancement. ROIs were drawn on enhancement DCE images and replicated to DCE parameter maps (Figure 2). Due to the. IP T. exploratory nature of this study, we selected these potential predictive DCE factors associated with outcome in ovarian cancer from a literature review of oncological MRI. Because of tumor biology complexity and scanty experience with DCE perfusion parameters in OC, mean, maximum, and minimum values were. SC R. registered from each DCE parameter for analyses. Furthermore, apparent diffusion coefficient (ADC) map was automatically generated from b-values of 0, 300, and 600mm2/s. For ADC analysis, the same slice. U. position and ROI placement were used as for the DCE analysis.. N. Statistical analysis. A. SPSS for Windows (Version 22.0, 2013, SPSS Inc., Chicago, IL, USA) was used for statistical analysis.. M. Values are presented as mean ± SD unless otherwise stated. Interclass correlation coefficient (ICC) was used. TE D. to test interobserver correlation, and the Bland–Altman method was applied to visualize interobserver variability. Kruskal–Wallis and Mann–Whitney U tests were used for classified parameters when appropriate, as were Spearman’s test for bivariate correlations for continuous variables, because of non-. EP. normal distribution of data. For the survival analyses, DCE parameters were dichotomized using the median as a cut-off. Recurrence-free survival (RFS) was defined as the interval between the date of surgery and the. CC. date of identified recurrence, and overall survival (OS) as the interval between the date of surgery and the date of death or the end of follow-up. The Kaplan–Meier method (log-rank) was used for univariate survival. A. analyses, and significant variables from univariate analyses were entered in a stepwise manner for Cox regression multivariate analysis. P ≤ 0.05 was considered significant, with high statistical significance set at P ≤ 0.01. Because of the exploratory nature of this study, we had decided to present p-values without the Bonferroni correction for multible comparisons.. Results. 7.

(10) A total of 38 patients were recruited. Three patients were excluded from DCE imaging analyses because of insufficient image quality: one for different scanning time and two for movement artefacts. Thus, 35 patients with primary OC (mean age 67 years, range 47–86 years) were included. We also performed subgroup analyses with stage III and IV patients who did not receive NACT (n = 24). One patient did not undergo cytoreductive debulking surgery after neoadjuvant chemotherapy due to massive tumor burden, stage IV. IP T. disease, large liver metastases that did not show response to chemotherapy, old age and poor performance status. Three other neoadjuvant patients had a good response to chemotherapy and they underwent debulking surgery. The mean largest solid tumor diameter in the plane where ROIs were placed was 75 mm (range 23–. SC R. 233 mm). Patients were followed up from the time of diagnosis until Septemper 2018.. U. Interobserver agreement was excellent for most of the DCE parameters (ICC, 0.951–0.994 for L-ROI; 0.928–0.991 for S-ROI) except for Vp and WashOut values, for which the two readers reached good. N. agreement (Table 3). The Bland-Altman method was used to visualize interobserver reproducibility (Figure. A. 3). The Bland–Altman 95% limits of agreement were -1.83 – 2.09 for Ktrans L-ROI and -1.70 – 2.22 for Ktrans. M. S-ROI, and coefficients of reproducibility were 0.27 and 0.53, respectively. Details of measurements for all. TE D. DCE parameters (mean values) are shown in Supplementary table 1.. Association between DCE parameters and OC prognostic factors. EP. Mean Ktrans values were higher in high-grade serous OC than other types (L-ROI, 0.913  0.952 vs. 0.425  0.262, P = 0.041; S-ROI, 1.001  1.051 vs. 0.436  0.333, P = 0.018). FIGO stage, dichotomized into two. CC. separate groups (FIGO 1+2 versus FIGO 3+4) was associated positively with Vp minimum (S-ROI, 2.29  1.84 vs. 5.89  4.67, P = 0.023). An optimal cytoreductive result from the operation (R0 vs. residual tumor). A. was associated with higher mean Ktrans (L-ROI, 0.874  0.535 vs. 0.755  1.140, P = 0.030; S-ROI, 0.967  0.538 vs. 0.810  1.330, P = 0.012; Figure 4). Platinum sensitivity was associated with higher WashIn (maximum L-ROI, 3.23  0.74 vs. 2.44  0.80, P = 0.016; maximum S-ROI, 3.13  0.81 vs. 2.39  0.82, P = 0.031) and higher WashOut (mean L-ROI, 0.180  0.090 vs. 0.111  0.062, P = 0.034) levels. Other DCE parameters showed no statistically significant associations with traditional prognostic factors. In subgroup. 8.

(11) analysis higher Ktrans remained significantly associated with high-grade serous OC (mean L-ROI, 1.08 ± 1.17 vs. 0.43 ± 0.30, P = 0.023; mean S-ROI, 1.21 ± 1.35 vs. 0.41 ± 0.33, P = 0.020) and with optimal cytoreductive result in staging surgery (mean L-ROI, 1.01 ± 0.43 vs. 0.78 ± 1.21, P = 0.020). Also higher Ve associated with high grade serous OC (mean S-ROI, 127.08 ± 119.82 vs. 40.17 ± 33.39, P = 0.010). Lower time-to-peak associated significantly (max S-ROI, 121.61 ± 124.09 vs. 186.04 ± 98.60, P = 0.030) with. IP T. Platinum sensitivity. Supplementary tables 2 and 3 show the detailed results.. DCE MRI parameters vs. ADC values. U. S-ROI, r = 0.366, P = 0.040), but not with other DCE parameters.. SC R. ADC value was correlated with DCE parameter time-to-peak (mean L-ROI, r = 0.387, P = 0.026; maximum. Recurrence free survival. N. In a median follow up of 57 months 15 of 35 patients experienced recurrence. The median recurrence free. A. survival (RFS) time was 19 months (range 6–68 months). In the univariate survival analysis DCE parameters. M. lower time-to-peak (maximum S-ROI, P = 0.047), higher Ve (minimum L-ROI, P = 0.035; maximum S-ROI,. TE D. P = 0.046) and higher Vp (maximum S-ROI, P = 0.033) correlated with shorter RFS in the Kaplan–Meier log-rank test (Figure 5). Other significant predictors of shorter RFS were advanced stage (P = 0.011), presence of residual tumor at surgery (P = 0.008), non-sensitivity to platinum-based chemotherapy (P <. EP. 0.001), and partial response to treatment (P < 0.001). We analysed survival data also for stage III and IV patients who did not receive NACT. In this 24 patient. CC. cohort Vp S-ROI minimum (P = 0.015), platinum resistance (P < 0.001) and progressive disease (P < 0.001) predicted shorter RFS.. A. To analyze whether multiparametric MRI (mpMRI) could even better predict the survival we made a combination variables of ADC and DCE parameters. Patients having both low ADC values and DCE parameters indicating more aggressive disease (higher in other DCE parameters but lower in time-to-peak) values were considered as ‘poor prognostic mpMRI’. In the cohort of 35 patients the compination of low ADC and high Ve, Vp, WashIn, WashOut, Peak or low time-to-peak velue proved to be significant predictors for shorter RFS. When we used only stage III and IV patients without NACT (n = 24 ) the combination of. 9.

(12) low ADC and high Ve, Vp, WashOut and Peak velue remained significant predictors for shorter RFS, see Table 4. In a multivariate analysis using Cox regression none of the factors remained significant.. Overall survival The median follow-up time was 38 months (range 2–90 months, two patients having died 2 months after. IP T. diagnosis). At the end of follow-up, 14 (40%) patients with OC had died. None of the DCE parameters were associated with survival in Kaplan–Meier analysis. In univariate survival analysis, the presence of residual tumor (P < 0.001), incomplete response to treatment (P < 0.001), and poor response to platinum-based. SC R. chemotherapy (P < 0.001) were significant predictors of poorer OS. Also the combination mpMRI variable with low ADC and high WashIn S-ROI minimum was significant (P = 0.033) (Table 4.). In the Cox. U. multivariate regression analysis, only response to treatment (P = 0.001, complete response, hazard ratio (HR) 1; partial response P = 0.002, HR 57.833, confidential interval (CI) 4.650-719.286; progressive disease P <. N. 0.001, HR 62.813, CI 6.792-580.899) proved to be an independent predictor of OS. We performed the. A. survival analysis also with the subgroup of stage III and IV patients without NACT (n = 24). DCE parameter. M. Time to Peak S-ROI maximum was significant (P = 0.037) in univariate survival analysis together with the. TE D. same clinical parameters as in the cohort of 35 patients. None of the mpMRI combination parameters predicted survival. In the Cox multivariate regression analysis only response to treatment remained significant (P = 0.005, complete response, HR 1; P = 0.002, partial response HR 81.963 (CI 5.091-. EP. 1319.485); P = 0.013 progressive disease HR 19.105 (CI 1.842-198.200)) although Time to Peak was nearly. CC. significant (P = 0.059, HR 3.556 (CI 0.954-13.253)).. A. Discussion. It is important to characterize ovarian masses as precisely as possible before treatment. We prospectively enrolled 38 patients with OC to explore whether semi-quantitative and pharmacokinetic perfusion DCE parameters are associated with the clinical course of OC and could thus be used as prognostic imaging biomarkers. Our results suggest that several DCE parameters are linked to advanced OC and can predict. 10.

(13) earlier recurrence. Also in the subgroup analyses with only stage III and IV patients several of the DCE parameters remained as significant prognostic factors.. Previous literature on correlations between DCE parameters and prognostic factors in other malignancies is promising. Many studies have shown that higher Ktrans, Kep, and Ve are associated with severity of. IP T. disease[8,21,24], while others have not found as clear correlations between prognostic factors and DCE parameters[25,26]. In contrast, in cervical cancer, low Ktrans was reported to be correlated with highly. metastatic tumors[19]. The results of the present exploratory study suggest that values indicative of higher. SC R. perfusion are associated to more aggressive OC.. In the present study, higher Ktrans was associated with high-grade serous OC, shown to be the aggressive type of OC[27]. Residual tumor is one of the predictors of poor survival in OC[28]. Interestingly, Ktrans correlated. U. with the presence of residual tumor on surgery: the lower the Ktrans, the bigger the risk for residual tumor in. N. cytoreductive surgery. According to clinical experience, surgery is more likely to succeed if the tumor is. A. highly vascular. With substantial fibrosis or a soft and fragile tumor, technical difficulties are more likely in. M. surgery[29,30]. There are no earlier studies comparing Ktrans values to cytoreductive surgical results.. TE D. Another important prognostic factor in OC is FIGO stage. Here, higher plasma volume fraction Vp correlated positively with FIGO stage, which is logical because tumors depend on angiogenesis and tumor neovascularization is strong in advanced stages[10,31]. Figo was dichotomized into FIGO 1 or 2 versus 3 or. EP. 4. Interestingly, in this cohort WashIn and WashOut parameters associated with platinum sensitivity, and WashOut remained significant also in the advanced stage subgroup. These findings need to be replicated in. CC. other cohorts. An imaging tool to estimate platinum sensitivity before treatment would be valuable. Shorter recurrence free survival was predicted by higher Ve and Vp and lower time-to-peak in our cohort.. A. Again, Vp reflects aggressive disease similar to low time-to-peak values. Earlier studies with perfusion parameters had discrepancies in predicting survival. Some studies found that a shorter enhanced time in the tumor predicted malignancy[6,11,13]. Other studies has showed that higher Ktrans and Kep corresponded with worse outcomes in breast cancer[24,32], while a study of cervical cancer showed that higher Ktrans was associated with better RFS[20]. A previous OC study found no correlation with patient progression-free interval and DCE measurements[18].. 11.

(14) We also created a multiparametric MRI variable combining low ADC and DCE parameters indicating more aggressive disease. In this analysis combinations of low ADC and high Ve, Vp, WashIn, WashOut, Peak or low time-to-peak parameters showed significant predictive value in the univariate survival analysis. Combinations of ADC and Ve, Vp, WashOut and Peak remained their signifigance also when the analysis was done only for the advanced stage subgroup (n = 24). Interestingly studies with multiparametric MRI in. IP T. other cancers have shown that a combination of results from different sequencies seems to be superior compared to analyzing each of them separately[33,34]. Conflicting results between DCE parameters and. prognostic factors in different studies may result partly from lack of consistent protocols for DCE studies and. SC R. partly from differing patient cohorts and nature and behavior of tumors.. The ADC values did not show correlations to most of the DCE parameters suggesting that these parameters. U. provide independent information on tumor biology. In general, lower ADC values are linked to high cellularity and solid tumors. Previous studies of DWI-MRI in OC support the view that lower ADC values. N. are associated with more aggressive disease[5,35]. However, a weak positive correlation between time-to-. A. peak and ADC values was found, giving support to an association of lower time-to-peak values with more. M. aggressive disease. In OC, no earlier studies have investigated the possible correlation between ADC and. TE D. time-to-peak. Sala et al. found a significant inverse correlation between pretreatment ADC and Kep values[36].. Because of the known heterogenic nature of OC[37] and the exploratory nature of our study, we also. EP. registered minimum and maximum values inside the ROI in addition to mean values to study how the heterogeneous voxel distribution affects the results. For example for grading gliomas, maximal perfusion. CC. value proved to be the most accurate[38]. Some previous studies have used histogram analyses to demonstrate this heterogenic distribution[21,39]. Similar to our results, earlier studies indicate that not all. A. measurements give significant results simultaneously; in the study by Kim et al., Ktrans50 was associated with tumor size, but Ktrans25 and Ktrans75 yielded no significant associations[21].. A main limitation of this study is the small sample size. Because of the relatively small population, the number of different histological subtypes is small and larger cohorts are needed to confirm the results. Although patients with high grade serous histological type were most frequent in our cohort, their. 12.

(15) distribution to different FIGO stages was relatively similar to other histological types (Table 1). Another limitation is that we used a free-hand technique for ROI placement on single slices; while three-dimensional voxel-by-voxel analyses might have yielded more reliable results for biological tumor heterogeneity. Our software did not allow histogram analyses so we used mean, maximum and minimum values to estimate heterogeneity. B1 correction was not performed. The acquisition time was 6.7sec/stack in perfusion scan for. IP T. 51 slices with an acquisition matrix of 269*387. AIF shape was inspected to be accurate for all patient individually. In further studies shorter temporal resolution may yield more robust data. When interpreting. results from DCE-MRI studies, it should be noted that the selected acquisition strategy and analysis protocol. SC R. affect the results, similar to type of contrast agent and data pre-processing. Only studies that use same type of scanners acquisition techniques and same analysis software are directly comparable.. U. A strength of our study is that we analyzed interobserver variability and obtained excellent ICC values for most DCE parameters and good ICC values for the remaining parameters. We aimed to explore the different. N. DCE parameters and prognostic factors in OC as widely as possible. Therefore pharmacokinetic perfusion. A. parameters and semi-quantitative measurements were included, although semi-quantitative values are not. TE D. M. directly comparable.. In conclusion, semi-quantitatively measured and pharmacokinetic DCE perfusion parameters revealed associations with the clinical course and prognostic factors of patients with OC. As Ktrans showed the highest. EP. associations with different prognostic factors and tumor aggressiveness, it may provide an imaging biomarker for future studies on OC. DCE parameters could also prove valuable in clinical practice and. CC. personalized medicine as Ktrans could predict success of debulking surgery and WashIn and WashOut could predict response to platinum based chemotherapy. In diagnostic imaging protocol of OC a combination of. A. tumor morphology, DWI and DCE parameters may be more useful than the single parameters. Further studies are needed to confirm our observations and clarify the prognostic value of DCE parameters in patients with OC.. Declaration of Interest statement: None. Authors does not have any conflicts of interest. 13.

(16) Conflict of Interest: The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.. IP T. Funding This study has received funding by Paavo Koistinen Foundation, Finnish adult education Foundation, Finnish Cultural Foundation, Kuopio University Hospital Research Foundation, Finnish Medical Foundation and. U. SC R. Kuopio University Hospital (VTR grant).. N. Acknowledgments. We thank Antti Lindgren, Pauli Vainio and Tuomas Selander for their skillful technical assistance. This. A. study was supported Paavo Koistinen Foundation, Finnish adult education Foundation, Finnish Cultural. TE D. University Hospital (VTR grant).. EP. Reference list [1]. M. Foundation, Kuopio University Hospital Research Foundation, Finnish Medical Foundation and Kuopio. J.A. Ledermann, F.A. Raja, C. Fotopoulou et al (2013) ESMO Guidelines Working Group, Newly. CC. diagnosed and relapsed epithelial ovarian carcinoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up., Ann. Oncol. 24:24-32. doi:10.1093/annonc/mdt333. J. Kaijser, T. Bourne, L. Valentin et al (2013) Improving strategies for diagnosing ovarian cancer: a. A. [2]. summary of the International Ovarian Tumor Analysis (IOTA) studies., Ultrasound Obstet. Gynecol. 41: 9–20. doi:10.1002/uog.12323.. [3]. S. Tingulstad, B. Hagen, F.E. Skjeldestad et al (1996) Evaluation of a risk of malignancy index based on serum CA125, ultrasound findings and menopausal status in the pre-operative diagnosis of pelvic masses., Br. J. Obstet. Gynaecol. 103: 826–31. http://www.ncbi.nlm.nih.gov/pubmed/8760716.. 14.

(17) [4]. K. Michielsen, R. Dresen, R. Vanslembrouck et al (2017) Diagnostic value of whole body diffusionweighted MRI compared to computed tomography for pre-operative assessment of patients suspected for ovarian cancer, Eur. J. Cancer. doi:10.1016/j.ejca.2017.06.010.. [5] [6]. J. Ren, Y. Huan, H. Wang et al (2008) Dynamic contrast-enhanced MRI of benign prostatic. IP T. hyperplasia and prostatic carcinoma: correlation with angiogenesis, Clin. Radiol. 63:153–159. doi:10.1016/j.crad.2007.07.023. [7]. V.B. Ho, S.F. Allen, M.N. Hood, P.L. Choyke (2002) Renal masses: quantitative assessment of. [8]. SC R. enhancement with dynamic MR imaging., Radiology. 224: 695–700. doi:10.1148/radiol.2243011048. H.R. Koo, N. Cho, I.C. Song et al (2012) Correlation of perfusion parameters on dynamic contrast-. U. enhanced MRI with prognostic factors and subtypes of breast cancers., J. Magn. Reson. Imaging. 36: 145–51. doi:10.1002/jmri.23635.. P.S. Tofts, B. Berkowitz, M.D. Schnall (1995) Quantitative Analysis of Dynamic Gd-DTPA. N. [9]. N. Paweletz, M. Knierim (1989) Tumor-related angiogenesis., Crit. Rev. Oncol. Hematol. 9: 197–. TE D. [10]. M. doi:10.1002/mrm.1910330416.. A. Enhancement in Breast Tumors Using a Permeability Model, Magn. Reson. Med. 33: 564–568.. 242. https://doi.org/10.1016/S1040-8428(89)80002-2. [11]. I. Thomassin-Naggara, M. Bazot, E. Daraï, et al (2008) Epithelial Ovarian Tumors: Value of. EP. Dynamic Contrast-enhanced MR Imaging and Correlation with Tumor Angiogenesis, Radiology. 248:148–159. doi:10.1148/radiol.2481071120. L. Bernardin, P. Dilks, S. Liyanage, et al (2012) Effectiveness of semi-quantitative multiphase. CC. [12]. dynamic contrast-enhanced MRI as a predictor of malignancy in complex adnexal masses:. A. radiological and pathological correlation., Eur. Radiol.22: 880–90. doi:10.1007/s00330-011-2331-z.. [13]. H.-M. Li, J.-W. Qiang, F.-H. Ma, S.-H. Zhao (2017) The value of dynamic contrast-enhanced MRI in characterizing complex ovarian tumors., J. Ovarian Res. 10: 4. doi:10.1186/s13048-017-0302-y.. [14]. I. Thomassin-Naggara, D. Balvay, E. Aubert, et al (2012) Quantitative dynamic contrast-enhanced MR imaging analysis of complex adnexal masses: a preliminary study, Eur. Radiol. 22:738–745. doi:10.1007/s00330-011-2329-6.. 15.

(18) [15]. R. Forstner, I. Thomassin-Naggara, T.M. Cunha, et al (2017) ESUR recommendations for MR imaging of the sonographically indeterminate adnexal mass: an update, Eur. Radiol. 27; 2248–2257. doi:10.1007/s00330-016-4600-3.. [16]. T. Taxt, R. Jirík, C.B. Rygh, et al (2012) Single-channel blind estimation of arterial input function and tissue impulse response in DCE-MRI., IEEE Trans. Biomed. Eng. 59: 1012–21.. [17]. IP T. doi:10.1109/TBME.2011.2182195. S.P. Li, A. Makris, M.J. Beresford, et al (2011) Use of Dynamic Contrast-enhanced MR Imaging to Predict Survival in Patients with Primary Breast Cancer Undergoing Neoadjuvant Chemotherapy,. [18]. SC R. Radiology. 260: 68–78. doi:10.1148/radiol.11102493.. C.L. Mitchell, J.P.B. O’Connor, A. Jackson, et al (2010) Identification of early predictive imaging. U. biomarkers and their relationship to serological angiogenic markers in patients with ovarian cancer with residual disease following cytotoxic therapy., Ann. Oncol. 21: 1982–9.. K.M. Øvrebø, C. Ellingsen, T. Hompland, E.K. Rofstad (2013) Dynamic contrast-enhanced magnetic. A. [19]. N. doi:10.1093/annonc/mdq079.. M. resonance imaging of the metastatic potential of tumors: A preclinical study of cervical carcinoma. [20]. TE D. and melanoma xenografts, Acta Oncol. (Madr). 52: 604–611.doi:10.3109/0284186X.2012.689851. B.R. Dickie, C.J. Rose, L.E. Kershaw, et al (2017) West, The prognostic value of dynamic contrastenhanced MRI contrast agent transfer constant K(trans) in cervical cancer is explained by plasma. [21]. EP. flow rather than vessel permeability., Br. J. Cancer. 116: 1436–1443. doi:10.1038/bjc.2017.121. S.H. Kim, H.S. Lee, B.J. Kang, et al (2016) Dynamic Contrast-Enhanced MRI Perfusion Parameters. CC. as Imaging Biomarkers of Angiogenesis, PLoS One. 11: e0168632. doi:10.1371/journal.pone.0168632. C.A. Cuenod, D. Balvay (2013) Perfusion and vascular permeability: Basic concepts and. A. [22]. measurement in DCE-CT and DCE-MRI, Diagn. Interv. Imaging. 94: 1187–1204. doi:10.1016/j.diii.2013.10.010.. [23]. H. Bickel, K. Pinker, S. Polanec, et al (2017) Diffusion-weighted imaging of breast lesions: Regionof-interest placement and different ADC parameters influence apparent diffusion coefficient values, Eur. Radiol.. doi:10.1007/s00330-016-4564-3.. 16.

(19) [24]. F. Liu, M. Wang, H. Li (2018) Role of perfusion parameters on DCE-MRI and ADC values on DWMRI for invasive ductal carcinoma at 3.0 Tesla, World J. Surg. Oncol. doi:10.1186/s12957-0181538-8.. [25]. H. Wright, J. Listinsky, C. Quinn, (2005) Increased ipsilateral whole breast vascularity as measured by contrast-enhanced magnetic resonance imaging in patients with breast cancer, in: Am. J. Surg.. [26]. IP T. doi:10.1016/j.amjsurg.2005.06.015. P. Andrzejewski, G. Wengert, T.H. Helbich, et al (2019) Sequential [18F]FDG-[18F]FMISO PET. and Multiparametric MRI at 3T for Insights into Breast Cancer Heterogeneity and Correlation with. SC R. Patient Outcomes: First Clinical Experience, Contrast Media Mol. Imaging. doi:10.1155/2019/1307247.. C.B. Gilks, D.N. Ionescu, S.E. Kalloger,et al (2008) Tumor cell type can be reproducibly diagnosed. U. [27]. and is of independent prognostic significance in patients with maximally debulked ovarian. A. du Bois, A. Reuss, E. Pujade-Lauraine, et al (2009) Role of surgical outcome as prognostic factor. A. [28]. N. carcinoma., Hum. Pathol. 39: 1239–51. doi:10.1016/j.humpath.2008.01.003.. M. in advanced epithelial ovarian cancer: A combined exploratory analysis of 3 prospectively. [29]. TE D. randomized phase 3 multicenter trials, Cancer. 115: 1234–1244. doi:10.1002/cncr.24149. R.K. Wong, V. Tandan, S. De Silva, A. Figueredo (2007) Pre-operative radiotherapy and curative surgery for the management of localized rectal carcinoma, in: Cochrane Database Syst. Rev.. [30]. EP. doi:10.1002/14651858.CD002102.pub2. K. Sato, S. Ito, T. Kitagawa, et al (2014) Factors affecting the technical difficulty and clinical. CC. outcome of endoscopic submucosal dissection for colorectal tumors, Surg. Endosc. 28: 2959–2965. doi:10.1007/s00464-014-3558-y. H.C. Hollingsworth, E.C. Kohn, S.M. Steinberg, et al (1995) Tumor angiogenesis in advanced stage. A. [31]. ovarian carcinoma., Am. J. Pathol. 147 33–41. http://www.ncbi.nlm.nih.gov/pubmed/7541612.. [32]. S.P. Li, A. Makris, M.J. Beresford, et al (2011) Use of dynamic contrast-enhanced MR imaging to predict survival in patients with primary breast cancer undergoing neoadjuvant chemotherapy, Radiology. 260:68–78. doi:10.1148/radiol.11102493. [33]. J.J. Park, C.K. Kim, S.Y. Park, et al (2014) Prostate cancer: Role of pretreatment multiparametric 3-T. 17.

(20) MRI in predicting biochemical recurrence after radical prostatectomy, Am. J. Roentgenol. doi:10.2214/AJR.13.11381. [34]. K. Pinker, L. Moy, E.J. Sutton, et al (2018) Diffusion-Weighted Imaging with Apparent Diffusion Coefficient Mapping for Breast Cancer Detection as a Stand-Alone Parameter: Comparison with Dynamic Contrast-Enhanced and Multiparametric Magnetic Resonance Imaging, Invest. Radiol.. [35]. IP T. doi:10.1097/RLI.0000000000000465. J.-W. Oh, S.E. Rha, S.N. Oh, et al (2015) Diffusion-weighted MRI of epithelial ovarian cancers:. correlation of apparent diffusion coefficient values with histologic grade and surgical stage., Eur. J.. [36]. SC R. Radiol. 84: 590–595. doi:10.1016/j.ejrad.2015.01.005.. E. Sala, M.Y. Kataoka, a. N. Priest, et al (2012) Advanced Ovarian Cancer: Multiparametric MR. U. Imaging Demonstrates Response- and Metastasis-specific Effects, Radiology. 263: 149–159. doi:10.1148/radiol.11110175.. H.A. Vargas, H. Veeraraghavan, M. Micco, et al (2017) A novel representation of inter-site tumour. N. [37]. A. heterogeneity from pre-treatment computed tomography textures classifies ovarian cancers by clinical. A. Hilario, A. Ramos, A. Perez-Nuñez, et al (2012) The added value of apparent diffusion coefficient. TE D. [38]. M. outcome, Eur. Radiol. 27:3991-4001. doi:10.1007/s00330-017-4779-y.. to cerebral blood volume in the preoperative grading of diffuse gliomas., AJNR. Am. J. Neuroradiol. 33: 701–7. doi:10.3174/ajnr.A2846.. F. Davnall, C.S.P. Yip, G. Ljungqvist, et al (2012) Assessment of tumor heterogeneity: an emerging. EP. [39]. A. CC. imaging tool for clinical practice? Insights Imaging 3:573-589. doi:10.1007/s13244-012-0196-6.. 18.

(21) IP T SC R U. N. Figure 1. Images from a 67-year-old woman with high-grade serous ovarian adenocarcinoma. A large pelvic. A. primary tumor is seen on T2-weighted (A), diffusion-weighted (DWIBS, b 800) (B) and dynamic contrast. M. enhancement (C, D) images. A large region of interest (L-ROI) is drawn to cover the whole tumor, excluding the necrotic area in the middle, and a small ROI (S-ROI) is placed on a subregion appearing to be the most. A. CC. EP. TE D. enhancing part of the tumor (D). The uterus has been marked with an asterisk.. 19.

(22) IP T SC R U N. A. Figure 2. Images from an 86-year-old woman with high grade (grade 3) serous ovarian adenocarcinoma. A. M. primary tumor is seen in (A) T2-weighted, (B) enhanced images, and (C) with color-encoded Ktrans (a rate constant for transfer of contrast agent from plasma to extravascular extracellular space (EES)) map, and (D). A. CC. EP. TE D. Kep (a rate constant for transfer of contrast agent from EES to plasma) map.. 20.

(23) IP T SC R U. N. Figure 3. Bland–Altman plots of Ktrans measurements obtained by two readers using (A) large regions of interest (L-ROIs) and (B) small regions of interest (S-ROIs). The difference in Ktrans values between the two. A. readers (y-axis) is plotted against the mean Ktrans of both readers (x-axis). The continuous line represents the. M. mean absolute difference (bias) in Ktrans between the two readers; the dashed lines represent the 95%. TE D. confidence intervals of the mean difference (limits of agreement). The mean absolute difference in DCE. A. CC. EP. Ktrans measurements between the two readers is higher when S-ROI is used.. 21.

(24) IP T SC R. Figure 4. The association between Ktrans levels and surgery results. Optimal cytoreductive results with. A. CC. EP. TE D. M. A. N. U. surgery (R0 or residual) were associated with higher mean Ktrans (L-ROI, P = 0.030; S-ROI, P = 0.012).. 22.

(25) IP T SC R U N A M TE D EP. CC. Figure 5. In Kaplan–Meier survival analysis, (A) a lower time-to-peak (maximum small region of interest (SROI): P = 0.047), (B) higher Ve (minimum large region of interest (L-ROI): P = 0.035), (C) maximum S-. A. ROI (P = 0.046), and (D) higher Vp (maximum S-ROI: P = 0.033) predict shorter recurrence-free survival (RFS).. 23.

(26) Tables Table 1. Clinicopathological characteristics of patients with ovarian cancer (n = 35).. A. CC. EP. TE D. M. A. N. U. SC R. IP T. Variable n (%) /median[range] Age years 67 [47-86] BMI 26 [17.4-40] CA-125 483 [16-5234] Stage at diagnosis I 5 (14.3) II 2 (5.7) III 14 (40) IV 14 (40) Histological type Serous high grade 23 (65.7) Endometrioid 5 (14.3) Mucinous 1 (2.9) Clear cell 1 (2.9) Other 5 (14.3) other Stage in high grade serous I 2 (8.7) 3 (25.0) II 1 (4.3) 1 (8.3) III 9 (39.1) 5 (41.7) IV 11 (47.8) 4 (25.0) Primary residual tumor a None 16 (47.1) ≤ 1 cm 14 (41.2) > 1 cm 4 (11.8) Chemotherapy response b Complete response 25 (71.4) Partial response 2 (5.7) Stable disease 0 Progressive disease 8 (22.9) c Tumor recurrence No recurrence 11 (40.7) Recurrence 16 (59.3) Platinum sensitivity Sensitive 25 (71.4) Resistant 10 (28.6) Patient status Dead, ovarian cancer 21 (60) Alive 14 (40) BMI = body mass index, a one patient did not enter to surgery because of overall situation, b four patient who received neoadjuvant chemotherapy and 3 dose of adjuvant chemotherapy after surgery are also included to final chemotherapy response analysis, c patients with progressive disease were not included to recurrence estimation. 24.

(27) Orient ation. TR (ms). TE (ms). Flip angle (°). FatSat. Resolution (mm). T2W_TSE 0:41.3 T2W_TSE 0:35.9 T2W_TSE 0:33.0 DWIBS 3:35.7 DWI_3b 3:40.6 dual_FFE 1:13.4. tra. Shortest. 80. 90. -. 0.7x0.7x5.0. sag. Shortest. 80. 90. -. cor. Shortest. 80. 90. -. tra. Shortest. Shortest. -. -. tra. Shortest. Shortest. -. STIR. tra. 180. 55. Dyn_eTH RIVE 2:38. tra. Shortest. 1.15 (outphase) 2.30 (inphase) Shortest. T1_FS_3D 0:20.2. tra. Shortest. Shortest. Other. 52 (0.5). 2.0. Breath hold. 0.7x0.7x5.0. 61 (0.5). 2.0. Breath hold. 0.7x0.7x5.0. 58 (0.5). 2.0. Breath hold. 62 (0). 2.0. b=800. 1.8x1.8x5.0. 56 (0.5). 2.0. -. 1.3x1.3x5.0. 56 (0.4). 2.0. b=0, 300, 600 Breath hold. 10. SPAIR. 0.9x0.9x5.0. 51 (0). 2.3. Dyn scan time 6.7sec. 10. SPAIR. 1.5x1.5x3.0. 147 (1.5). 2.0. Breath hold. U. 1.3x1.3x5.0. N. M. TE D. N slices (gap mm). SC R. SENSE factor. A. Sequence acquisition time. IP T. Table 2. Imaging protocol. A. CC. EP. TR = repetition time, TE = echo time, FatSat = fat saturation, N slices = number of slices, tra = transversal, sag = sagittal, cor = coronal, TSE = turbo spin echo, DWIBS = diffusion-weighted imaging with background body signal suppression, STIR = short tau inversion recovery, SPAIR = spectral attenuated inversion recover, FFE = fast field echo, Dyn_eTHIRIVE = Dynamic contrast-enhanced T1 high-resolution isotropic volumetric examination. 25.

(28) Table 3. Interclass correlations (ICCs) between two readers from all the DCE parameters (mean values) used in analyses, both large regions of interest (L-ROIs) and small regions of interest (S-ROIs). ICC L-ROI. ICC S-ROI. Ktrans. 0.994. 0.981. Kep. 0.990. 0.952. Ve. 0.997. 0.965. Vp. 0.637. 0.614. AUC. 0.991. 0.991. Peak. 0.971. 0.944. Time to peak. 0.975. 0.790. WashIn. 0.969. 0.928. WashOut. 0.951. 0.584. M. A. N. U. SC R. IP T. DCE parameter. A. CC. EP. TE D. AUC = area under the curve; ICC = Interclass correlation; L-ROI = large region of interest; S-ROI = small region of interest. 26.

(29) 27. EP. CC. A TE D. IP T. SC R. U. N. A. M.

(30) Table 4. Univariate survival analysis (p ≤ 0.05 considered significant) with multiparametric MRI variables. Multiparametric MRI parameter combined low ADC values (L-ROI < 0.821 and S-ROI < 0.688) and DCE parameters indicating more aggressive disease (higher than median in other DCE parameters but lower than median in time-to-peak). RFS all 35 patients 0.002. RFS 24 patients 0.012. OS all 35 patients ns. OS 24 patients. A. CC. EP. TE D. M. A. N. U. SC R. IP T. Ve L-ROI min + low ns ADC Ve S-ROI max + low 0.002 0.020 ns ns ADC Ve S-ROI min + low 0.001 0.005 ns ns ADC Vp L-ROI min + low <0.001 0.005 ns ns ADC Vp S-ROI mean + low 0.009 ns ns ns ADC Vp S-ROI max + low 0.005 ns ns ns ADC WashIn S-ROI min + 0.024 ns 0.033 ns low ADC WashOut L-ROI mean ns 0.027 ns ns + low ADC WashOut S-ROI min + 0.048 ns ns ns low ADC Peak S-ROI mean + 0.001 0.005 ns ns low ADC Peak S-ROI max + low ns 0.012 ns ns ADC Time-to-peak S-ROI 0.050 ns ns ns mean + low ADC Time-to-peak S-ROI 0.007 ns ns ns max + low ADC RFS = recurrence free survival, OS = overall survival, 24 patients were stage III and IV without neoadjuvant chemotherapy, S-ROI = Small ROI, L-ROI = Large ROI, ns = not significant. 28.

(31) Supplementary table 1. Detailed calculations of Bland–Altman variables, Bland–Altman 95% limits of agreement, and coefficients of reproducibility for all DCE parameters used in analyses, both for large region of interest (L-ROI) and small region of interest (S-ROI). DCE parameter BA 95% L-ROI BA 95% S-ROI CR L-ROI CR S-ROI -1.83 to 2.09. -1.70 to 2.22. 0.27. 0.53. Kep. -1.82 to 2.10. -1.63 to 2.29. 0.28. 0.64. Ve. 6.30–10.22. 30.73–34.65. 16.68. 67.61. Vp. 11.67–15.59. 13.82–17.74. 26.53. 30.57. AUC. 960.8–964.7. 1143.5–1147.4. 2010.27. Peak. 6.13–10.05. 10.02–13.94. 17.47. Time to peak. 7.84–11.76. 30.39–34.31. 18.97. WashIn. -1.65 to 2.27. -1.44 to 2.48. 0.63. WashOut. -1.92 to 2.00. -1.78 to 2.14. 0.08. IP T. Ktrans. 2483.73. U. SC R. 27.14 62.70 1.06 0.36. A. CC. EP. TE D. M. A. N. AUC = area under the curve; BA 95% = Bland–Altman 95% limits of agreement; CR = coefficient of repeatability. 29.

(32) Supplemantary table 2. Dynamic contrast enhanced perfusion parameters association with clinical. 0.766. 0.204. 0.290. 0.275. 0.457. 0.186. 0.545. 0.018. 0.201. 0.596. .094. 0.290. 0.568. 0.654. 0.321. U. 0.963. 0.205. 0.898. 0.799. 0.902. 0.317. 0.532. 0.535. 0.596. 0.828. 0.900. 0.560. 0.308. 0.371. 0.386. 0.756. 0.717. 0.183. 0.061. 0.880. 0.918. 0.134. N. 0.168. M. 0.837. 0.955. 0.947. 0.360. 0.517. 0.313. 0.940. 0.809. 0.974. 0.957. 0.885. 0.503. 0.237. 0.711. 0.112. 0.251. 0.074. 0.741. 0.490. 0.577. 0.604. 0.638. 0.421. 0.407. 0.083. 0.257. 0.711. 0.821. 0.655. 0.564. 0.387. 0.983. 0.084. 0.097. 0.082. 0.605. 0.492. 0.482. 0.666. 0.829. 0.473. 0.581. Ve S-R min 0.076. 0.416. 0.328. 0.157. 0.224. 1.00. Vp L-R mean. 0.146. 0.280. 0.589. 0.266. 0.567. CC. 0.241. Ve L-R min. A. Ve S-R mean Ve S-R max. 0.571. ADC unclassifie d. 0.610, r0.087 0.969, r0.007 0.494, r0.162 0.796, r0.045 0.953, r0.010 0.994, r0.014. 0.625, r0.088 0.859, r0.032 0.803, r0.061 0.520, r0.116 0.708, r0.068 0.491, r0.141. 0.565, r0.101 0.794, r0.046 0.841, r0.053 0.127, r0.267 0.202, r0.224 0.293, r0.214. 0.560, r0.105 0.653, r0.081 0.922, r0.026 0.720, r0.066 0.976, r0.005 0.194, r0.275. 0.809, r0.042 0.889, r0.025 0.808, r0.064 0.789, r0.048 0.786, r0.048 0.922, r0.020. 0.812, r0.043 0.593, r0.097 0.749, r0.058 0.987,r0.003 0.556, r0.108 0.878, r0.033. 0.660, r0.078. 0.364, r0.166. IP T. 0.134. CA-125 unclassif ied. SC R. 0.041. TE D. Ve L-R mean Ve L-R max. age. 0.853. EP. Kep L-R mean Kep L-R max Kep L-R min Kep S-R mean Kep S-R max Kep S-R min. HGSO platinum C or sensitivity other. A. variables, all 35 patient are included Gradus FIGO Residua 12-34 in surgery 0 or more trans K L-R 0.197 0.592 0.030 mean Ktrans L-R 0.172 0.483 0.112 max Ktrans L-R 0.211 0.186 0.804 min Ktrans S-R 0.121 0.805 0.012 mean Ktrans S-R 0.209 0.483 0.091 max Ktrans S-R 0.220 0.341 0.497 min. 30.

(33) 1.00. 0.891. 0.615. 1.00. 1.00. 0.121. 1.00. 1.00. 0.701. 0.159. 0.564. 0.428. 0.133. 0.435. 0.801. 0.481. 0.449. 0.943. 0.711. 0.931. Vp S-R min 0.327. 0.023. 0.619. 0.368. 0.105. 0.895. 0.378. 0.364. 0.408. 0.862. 0.381. 0.487. 0.511. 0.386. 0.157. 0.835. 0.165. 0.619. 0.206. 0.271. 0.309. 0.492. 0.196. 0.383. 0.680. 0.581. 0.578. 0.511. 0.393. 0.364. 0.370. 0.889. 0.273. 0.820. 0.837. 0.809. 0.297. 0.742. 0.293. 0.265. 0.388. 0.109. 0.161. 0.202. 0.354 0.508 0.197. 0.228. 0.596. 0.768. 0.116. 0.716. 0.186. 0.366. 0.633. 0.072. 0.540. 0.650. 0.809. 0.375. 0.798. 0.320. 0.265. EP. 0.512. 0.821. 0.342. 0.691. 0.936. 0.902. 0.704. 0.181. 0.942. 0.220. 0.349. 0.187. 0.355. 0.498. 0.609. 0.175. 0.668. 0.741. 0.679. 0.297. 0.333. 0.398. 0.393. 0.692. 0.243. 0.059. 1.00. 0.028. 0.729. 0.058. 0.352. 0.768. 0.742. 0.843. 0.835. 0.127. 0.756. 0.509. 0.221. 0.214. A. TtP L-R mean TtP L-R max TtP L-R min TtP S-R mean TtP S-R max. 0.259. 0.903. CC. Peak L-R mean Peak L-R max Peak L-R min Peak S-R mean Peak S-R max Peak S-R min. 0.368. U. AUC L-R mean AUC L-R max AUC L-R min AUC S-R mean AUC S-R max AUC S-R min. TE D. Vp S-R mean Vp S-R max. A. Vp L-R min 0.283. 0.291 0.103. 0.915, r0.019 0.683, r0.159 0.586, r0.097 0.919, r0.018 0.917, r0.024. 0.277, r0.198 0.854, r0.098 0.779, r0.119 0.326, r0.179 0.729, r0.083. IP T. 0.943. N. 0.431. 0.352, r0.162 0.322, r0.173 0.235, r0.213 0.490, r0.121 0.258, r0.196 0.491, r0.120. 0.217, r0.221 0.208, r0.225 0.325, r0.183 0.444, r0.138 0.437, r0.140 0.236, r0.212. 0.285, r0.186 0.157, r0.244 0.148, r0.257 0.498, r0.118 0.124, r0.265 0.683, r0.071. 0.147, r0.258 0.124, r0.273 0.329, r0.181 0.246, r0.208 0.167, r0.246 0.191, r0.233. 0.640, r0.082 0.701, r0.067 0.593, r0.096 0.264, r0.194 0.063, r0.318. 0.026, r0.387 0.960, r0.009 0.406, r0.155 0.142, r0.261 0.040, r0.366. SC R. 0.355. M. Vp L-R max. 31.

(34) 0.195, r0.235 0.191, r0.237 0.779, r0.119 0.364, r0.166 0.632, r0.088 0.674, r0.088. 0.459, r0.129 0.920, r0.018 0.493, r0.120 0.730. r0.060 0.388, r0.151 0.701, r0.139. 0.877, r0.028 0.607, r0.093 0.734, r0.066 0.921, r0.020 0.910, r0.020 0.957, r0.010. 0.942. 0.716. 0.513. 0.142. 0.090. 0.408. 0.059. 0.490. 0.413. 0.297. 0.066. 0.311. 0.016. 0.469. 0.439. 0.121. 1.00. 0.462. 0.558. 0.197. 0.639. 0.085. 0.183. 0.277. 0.104. 0.730. 0.517. 0.259. 0.313. 0.519. 0.031. 0.334. 0.680. 0.658. 0.884. 0.132. 0.825. 1.00. 0.327. 0.621. 0.427. 0.555. 0.034. 0.371. 1.00. 0.138. 0.314. 0.154. 0.314. 0.216. 0.730. 0.917. 0.214. 0.196. 0.837. 0.679. 0.532. 0.443. 0.691. 0.116. 0.741. 0.490. 0.445. 0.661. 0.643. 0.606. 0.384. 0.221. 0.068. 0.136. 0.732. 0.529. U. 0.596 0.947. IP T. 0.876. SC R. 0.522, r0.114 0.597, r0.094 0.356, r0.350 0.221, r0.216 0.415, r0.144 0.697, r0.078. 0.756. N. WashOut L-R mean WashOut L-R max WashOut L-R min WashOut S-R mean WashOut S-R max WashOut S-R min. 0.363, r0.164. 0.216. A. WashIn L-R mean WashIn L-R max WashIn L-R min WashIn SR mean WashIn SR max WashIn SR min. 0.419, r0.141. 0.569. M. TtP S-R min. A. CC. EP. TE D. FIGO = International Federation of Gynecology and Obstetrics stage, HGSOC = high grade serous ovarian cancer, ADC = apparent diffusion coefficient, L-R = Large region of interest, S-R = Small region of interest, TtP = time-to-peak. 32.

(35) Supplemantary table 3. Dynamic contrast enhanced perfusion parameters association with clinical. CA-125 unclassified. ADC unclassified. 0.639. 0.816, r0.050 0.955, r0.012 0.067, r0.522 0.829, r0.047 0.953, r0.013 0.074, r0.371. 0.875, r0.034 0.799, r0.055 0.734, r0.104 0.704, r0.082 0.613, r0.109 0.489, r0.174. 0.907 0.886 0.292 0.953 0.757. Kep L-R mean Kep L-R max Kep L-R min Kep S-R mean Kep S-R max Kep S-R min. 0.481. 0.142. 0.284. 0.374. 0.903. 0.327. 0.540. 0.270. 0.465. 0.855. 0.540. 0.206. 0.519. 0.121. 0.107. 0.272. 0.620. 0.949. 0.272. 0.478. 0.272. 0.804. Ve L-R mean Ve L-R max. 0.178. 0.715. 1.00. 0.910. 0.178. 0.066. 0.043. 0.126. 0.682. EP. 0.626. 0.667. 0.580. 0.580. 0.623. 0.814. 0.361. 0.584. 0.414. 0.527. 1.00. Ve S-R mean Ve S-R max Ve S-R min. 0.519. 0.093. 0.010. 0.175. 0.321. 0.747. 0.519. 0.197. 0.606. 0.951. 0.143. 0.172. 0.073. 0.497. 0.501. Vp L-R mean. 0.138. 0.796. 0.796. 0.316. 0.659. A. Ve L-R min. 0.884, r0.031 0.321, r0.211 0.670, r0145 0.355, r0.202 0.890, r0.031 0.554, r0.154. 0.494, r0.147 0.610, r0.110 0.790, r0.091 0.812, r0.052 0.647, r0.101 0.209, r0.321. 0.578, r0.120 0.910, r0.024 0.285, r0.355 0.497, r0.149 0.754, r0.069 0.808, r0.064. 0.990, r0.003 0.336, r0.205 0.5041, r0.143 0.667, r0.095 0.356, r0.202 0.844, r0.051. 0.433, r0.172. 0.381, r0.192. U 0.815. N. A. M. TE D. 0.770. CC. 0.118. IP T. age. SC R. variables, 24 patient with advance stage are included. Gradus Residua HGSOC platinum in or other sensitivity surgery 0 or more Ktrans L-R 0.058 0.142 0.014 0.023 mean Ktrans L-R 0.086 0.098 0.086 0.298 max Ktrans L-R 0.475 0.668 0.866 0.324 min Ktrans S-R 0.076 0.221 0.010 0.020 mean Ktrans S-R 0.098 0.066 0.066 0.298 max Ktrans S-R 0.021 0.258 0.183 0.524 min. 33.

(36) 0.478. 0.606. 0.220. 0.947. 0.900. 0.180. 0.180. 0.651. 0.655. 0.655. 0.333. 0.651. 0.897. 0.256. 0.313. 0.949. 0.796. 0.401. 0.738. 0.529. 0.322. 0.066. 0.586. 0.066. 0.565. AUC L-R mean AUC L-R max AUC L-R min AUC S-R mean AUC S-R max AUC S-R min. 0.624. 0.540. 0.903. 0.501. 0.143. 0.624. 0.245. 0.426. 0.358. 0.061. 0.275. 0.682. 0.378. 0.453. 0.210. 0.391. 0.501. 0.582. 0.582. 0.046. 0.327. 0.540. 0.540. 0.501. U. 0.624. 0.582. 0.624. 0.903. Peak L-R mean Peak L-R max Peak L-R min Peak S-R mean Peak S-R max Peak S-R min. 0.561. 0.581. 0.878. 0.540. 0.219. 0.133. 0.462. 0.783. 0.540. 0.079. TtP L-R mean TtP L-R max TtP L-R min TtP S-R mean. M. 0.164, r0.300 0.800, r0.200 0.893, r0.030 0.235, r0.258 0.887, r0.042 0.235, r0.252 0.440, r0.165 0.301, 0.231 0.277, r0.231 0.357, r0.197 0.500, r0.148. SC R. A. N. 0.040 0.160. 0.733. 0.622. 0.275. 0.262. 0.713. 0.444. 0.854. 0.143. 0.545, r0.130 0.543, r0.130 0.172, r0.302 0.578, r0.120 0.318, r0.213 0.829, r0.047. 0.213, r0.264 0.538, r0.132 0.359, r0.206 0.204, r0.269 0.206, r0.268 0.291, r0.225 0.151, r0.303 0.670, r0.092 0.842, r0.045 0.243, r0.248. 0.105. 0.668. 0.691. 0.903. 0.114. 0.830. 0.426. 0.374. 0.951. 0.241. 0.561. 0.668. 0.561. 0.066. 0.770. 0.066. 0.408. 0.426. 0.098. 0.690. 0.838. 0.411. 0.517. 0.152. 0.057. 0.733. 0.003. 0.273. 0.603. 0.903. 0.646. 0.076. 0.953. 0.131. CC. A. 0.453, r0.161 0.689, r0.086 0.225, r0.269 0.576, r0.120 0.578, r0.120 0.323, r0.211. EP. 0.209. TE D. 0.657. 0.781, r0.061 0.600, r0.400 0.274, r0.238 0.497, 0.149 0.573, r0.165. IP T. Vp L-R max Vp L-R min Vp S-R mean Vp S-R max Vp S-R min. 34.

(37) 0.540. 0.358. 0.030. 0.187. 0.038. 0.209. 0.854. 0.878. 0.298. 0.861. 0.431. WashIn LR mean WashIn LR max WashIn LR min WashIn SR mean WashIn SR max WashIn SR min. 0.519. 0.156. 0.204. 0.366. 0.901. 0.366. 0.121. 0.082. 0.071. 0.664. 0.827. 0.513. 0.275. 0.643. 0.380. 0.747. 0.401. 0.082. 0.747. 0.710. 0.401. 0.606. 0.204. 0.175. 0.951. 0.922. 0.920. 0.132. 0.546. 0.847. 0.923, r0.021 0.875, r0.035 0.672, r0.093 0.738, r0.074 0.998, r0.000 0.554, r0.154. WashOut L-R mean WashOut L-R max WashOut L-R min WashOut S-R mean WashOut S-R max WashOut S-R min. 0.298. 0.221. 0.142. 0.076. 0.320. 0.142. 0.142. 0.126. 0.178. 0.245. 0.391. 0.198. 0.270. 0.178. 0.903. 0.759. 0.327. 0.861. 0.142. 0.426. 0.624. 0.903. 0.815. 0.275. 0.096. 0.101. 0.739. 0.670, r0.092 0.804, r0.053 0.602, r0.112 0.465, r0.157 0.392, r0.183 0.693, r0.167. U 0.482. N. A. M. TE D. 0.317. 0.122, r0.324 0.257, r0.076 0.391, 0.188 0.599, r0.116 0.957, r0.029 0.631, r0.106 0.996, r0.001 0.729, r0.091. IP T. 0.481. SC R. TtP S-R max TtP S-R min. 0.412. 0.695, r0.084 0.885, r0.031 0.590, r0.116 0.872, r0.035 0.585, r0.117 0.750, r0.069. A. CC. EP. HGSOC = high grade serous ovarian cancer, CA-125 = cancer antigen-125, ADC = apparent diffusion coefficient, L-R = Large region of interest, S-R = Small region of interest, TtP = time-to-peak. 35.

(38)