Dynamic modeling of uteroplacental blood flow in IUGR indicates vortices and elevated pressure in the intervillous space – a pilot study

Ischemic placental disease is a concept that links intrauterine growth retardation (IUGR) and preeclampsia (PE) back to insufficient remodeling of uterine spiral arteries. The rheological consequences of insufficient remodeling of uterine spiral arteries were hypothesized to mediate the considerably later manifestation of obstetric disease. However, the micro-rheology in the intervillous space (IVS) cannot be examined clinically and rheological animal models of the human IVS do not exist. Thus, an in silico approach was implemented to provide in vivo inaccessible data. The morphology of a spiral artery and the inflow region of the IVS were three-dimensionally reconstructed to provide a morphological stage for the simulations. Advanced high-end supercomputing resources were used to provide blood flow simulations at high spatial resolution. Our simulations revealed turbulent blood flow (high-velocity jets and vortices) combined with elevated blood pressure in the IVS and increased wall shear stress at the villous surface in conjunction with insufficient spiral artery remodeling only. Post-hoc histological analysis of uterine veins showed evidence of increased trophoblast shedding in an IUGR placenta. Our data support that rheological alteration in the IVS is a relevant mechanism linking ischemic placental disease to altered structural integrity and function of the placenta.

2 system (4x3 XY), automated slide loader with proScan III stage (Prior Scientific, Jena, Germany) and color digital camera (2/3 CCD chip 1,4 MP, 1388x1049 pixel; MBF Bioscience, Williston, VT, USA), and focus encoder (MBF Bioscience). The entire system was controlled by the software Stereo Investigator (MBF Bioscience). Then, the scan files were aligned with Amira software (version 5.5.0; Visualization Sciences Group, FEI Company, Burlington, MA, USA). All images (size: 2560x1518 pixels; resolution: 5.9 µm/pixel, section thickness: 10 µm; distance between sections: 50 µm) were converted to DICOM and stacked to generate a three-dimensional data set ( Figure S1B). From these data, blood was segmented in the region of the uterine spiral artery (marked red in Figure S1B), and the intervillous space (IVS) at the opening of the artery was marked yellow in Figure S1B using Mimics software (version 16.0; Materialise, Leuven, Belgium). The segmented geometry was closed using symmetry conditions to account for a small part of the artery that was not contained in the placenta sample (using Mimics software).
Additionally, a second remodeled artery geometry was created by dilating the outlet radius by a factor of 3 within a smooth transition zone in the last 3 mm of the artery length according to [12] (using Mimics software). The resulting geometry of both arteries was connected to the IVS, improved in quality [46] and exported in STL file format. The final arterial length was approximately 10 mm with a mean inlet diameter of 0.34 mm and an outlet diameter of 1.15 mm for the clinically normal remodeled uterine spiral artery, and 0.39 mm for the pathological artery, which corresponds well to previous observations [12,47]. The IVS proximal to the artery opening and included in the stage (proximal IVS) had dimensions of 3.0x2.5x2.0 mm and was approximately the size that is supplied with blood via a single uterine spiral artery [12]. A threedimensional tetrahedral mesh was generated using Gmsh (version 2.9.3; freeware by [48]) and consisted of 1,548,144 tetrahedral elements in the clinically normal case and 1,502,567 elements in the pathological cases.
Tissue processing for post-hoc histological analysis. For post-hoc microscopic analysis of trophoblast shedding, placentas of a clinically normal pregnancy and an IUGR pregnancy were used (Table S1). These placentas were processed for histology in an innovative, non-routine manner. To this end the placentas were fixed in 4.5% phosphate-buffered formaldehyde (Roti-Histofix, Carl Roth, Karlsruhe, Germany) for 7 days and embedded in paraffin (Paraplast, melting point 56−58°C; Leica No. 39602012; Wetzlar, Germany), each placenta without tissue sampling as a whole. After fixation, embedding began with passage through a serial stepwise ethanol gradient 3 from 50% ethanol for 6 h, 70% ethanol for 6 h, second time 70% ethanol for 12 h, 80% ethanol for 12 h, 96% ethanol for 12 h, and two times 100% ethanol for 15 h. From pure ethanol, the placentas were transferred in xylene (two times in xylene for 12 h and 8.5 h). From xylene, the placentas were transferred to paraffin at 65°C via three infiltration steps for 24 h each. A special form with a diameter of 32 cm was used for the entire embedding procedure. To enable sectioning, the paraffin block containing the whole placenta was cut into four quarters. Each quarter block was oriented for sectioning such that sections were taken in a plane parallel to the chorionic surface, starting at the basal plate of the placenta (Microtom, Polycut S; Reichert Jung, Wetzlar, Germany). From the level of the first appearance of tissue in the section onwards, serial sections with a thickness of 20 µm were prepared and collected between numbered sheets of paper. All sections were mounted on oversize object slides (120x150x1 mm). For standard hematoxylin-eosin (HE), the object slides were coated with 0.5% alum chrome gelatin (7.5  For immunohistochemistry, the object slides were coated with poly-L-lysine (Sigma, article No. P8920, Munich, Germany). For coating with poly L-lysine, the object slides were degreased in acetone for 10 min, incubated with poly-L-lysine (1:10 in distilled water) for 30 min and dried at 60°C for 1 h. Prior to tissue staining or immunohistochemical processing, the mounted paraffin sections were deparaffinized by transferring from xylene (two times for 10 min and 5 min in xylene) through a descending ethanol series (100% ethanol two times for 3 min, 96% ethanol for 3 min, 80% ethanol for 3 min, 70% ethanol for 3 min, 50% ethanol for 3 min, and finally distilled water for 3 min). Every 10th section was stained with standard hematoxylin-eosin (HE) and the following serial section by immunohistochemistry (see below). Trophoblast nature of cellular elements in the sections was ensured by using anti-cytokeratin 7 (CK7, Dako Clone OV TL12/30 Code M7018, monoclonal mouse; Santa Clara, CA, USA) reactivity. These values are in good agreement with values reported in [18,37]. 5 Assuming that this total blood flow provided to the placenta distributes equally to a number of n=130 spiral arteries, which is the average of the values provided in the literature (n=60 in [12] and n=200 in [18]), the blood flow through a single uterine spiral artery reads as  Q  (Q AU,l  Q AU,r ) n and was used to feed the simulations of blood flow in this study (Figure 3 in the main text).
Compared to [12], our values for maximum blood flow Q through a single spiral artery (Table S2) are conservative due to the larger number of spiral arteries assumed in our models [12].
The full time span of the three visualized waveforms (normal, IUGR, and IUGR/PE) during the three maternal heart cycles was simulated in each case, which led to the following simulation setups: (i) clinically normal situation: dilated spiral artery geometry with clinically normal waveform; (ii) IUGR: un-dilated spiral artery geometry with IUGR waveform; and (iii) IURG/PE: un-dilated spiral artery geometry with IUGR/PE waveform. During the first maternal heart cycle, flow was allowed to fully develop from the initial resting state, and the second maternal heart cycle provided representative rheological properties in the spiral artery. Therefore, the estimates of the models were all taken from the values of the 2 nd maternal heart cycle, i.e., the interval of 1.2 to 2.0 s. This interval was also the basis for the figures and movies and is indicated by the black bar in Figure 3 in the main text.
6 Tables   Table S1. Macroscopic clinical data of the placentas that were processed for histology. This includes the tissue sample that provided the morphological stage of the present study (threedimensional [3D] reconstruction of a spiral artery and the corresponding proximal IVS of an IUGR placenta [see Fig.S2]) and the two placentas that were non-routinely processed post-hoc to analyze trophoblast shedding (GA, gestational age; BW, birth weight; PW, placental weight; PW/BW, placental birth weight ratio; LD, longest diameter of the placenta; SD, shortest diameter of the placenta; surface area of the placenta; roundness of the placenta; thickness of the placenta).  Figure S1.