Erythrocyte sedimentation rate as a new diagnostic biomarker for neuroacanthocytosis syndromes

Chorea-acanthocytosis (ChAc) and McLeod syndrome (MLS) are the core diseases among the group of rare neurodegenerative disorders that comprise neuroacanthocytosis syndrome (NAS). Both ChAc and MLS patients present with an increased number of irregularly spiky erythrocytes, so-called acanthocytes. The detection of acanthocytes is often a crucial parameter in the diagnosis of NAS. However, this approach is error-prone and not very reliable, typically explaining the misdiagnosis of NAS patients. Based on the standard Westergren method, we show that compared with that in healthy controls, the erythrocyte sedimentation rate (ESR) with a two-hour read-out is significantly prolonged in ChAc and MLS with no overlap. Thus, the ESR is a clear, robust and easily obtained diagnostic marker. Mechanistically, by applying modern colloidal physics, we show that acanthocyte aggregation and plasma fibrinogen levels slow the sedimentation process. Apart from its diagnostic value, ESR may also be the first biomarker for monitoring treatments for NAS patients. Further studies are required to test whether the ESR may also detect other NASs. In addition to medical progress, this study is also a hallmark of the physical view of the erythrocyte sedimentation process by describing anticoagulated blood in stasis as a percolating gel, allowing the application of colloidal physics theory.


Introduction
1 and often result in severe disability, while treatment options 14 remain purely symptomatic so far 9 . Diagnosis is confirmed 15 by genetic testing (VPS13A gene in ChAc, XK gene in MLS) 16 and chorein in Western blot (ChAc) or immunohematological 17 assessment (MLS) 5 , although the initial detection often relies 18 on blood smears 10 . One of the common features of NAS is the 19 presence of acanthocytes among erythrocytes 4-6 . Acantho-20 cytes are deformed erythrocytes whose irregular membrane 21 presents disordered, asymmetric spikes. 22 For many of these rare diseases, diagnosis is challenging, 23 and misdiagnosis is frequent, which additionally increases the 24 patients' burden of disease. One of the reasons for delayed 25 diagnosis is related to the difficult detection of acanthocytes 26 in classical blood smears and routine laboratory testing: sen-27 tubes, the time curves of representative ESR measurements,  Figure 2B shows a number of representative acanthocytes in 78 3D, which were used for further analysis. 79 We observed that the ESR tends to decrease when the frac-80 tion of acanthocytes increases. To relate the ESR to the num-81 ber of acanthocytes, the inverse of the ESR at 2 h is plotted 82 as a function of the number of acanthocytes (Fig. 2C). In a 83 more general approach, we plotted the inverse of the ESR 84 at 2 h against the number of all deformed cells (echinocytes, 85 acanthocytes and others; Fig. 2D). The analysis revealed a 86 significant correlation for both parameters (R 2 coefficient of 87 0.61 and 0.59, respectively). Consistent with this finding, 88 the p-value decreased from 0.037 (acanthocytes) to 0.011 (all 89 deformed cells). 91 Because there was no significant difference in the ESR be-92 tween ChAc and MLS patients (Fig. 1), we do not discrimi-93 nate between ChAc and MLS patients for further statistical 94 comparisons. Therefore, we subsume them into one NAS 95 group. 96 The reasons for the differences in the ESR are varied and 97 can depend on the erythrocytes, the blood plasma or the 98 erythrocytes-plasma volume relation (i.e. the hematocrit or 99 erythrocyte volume fraction φ ). The last reason can be ex-100 cluded since there was no significant difference in the hema-101 tocrit between the groups (cp. Table 1 and Supplemental 102 Fig. S1B). The primary physical parameters such as the den-103 sity of the individual erythrocytes, represented by the mean 104 cellular hemoglobin concentration (MCHC), as well as the 105 constitution of the plasma can also be part of the explanation. 106 However, the MCHC indicates a higher density of erythro-107 cytes in the NAS patients (Supplemental Fig. S3C), a feature 108 that would favor faster sedimentation and therefore rules it 109 out as an explanation. In contrast, the total plasma protein 110 analysis (Supplemental Fig. S4A), as a representative of the 111 plasma content, indicates a lower protein concentration and 112 therefore a first hint for a plasma contribution to the ESR 113 change. In the following sections, we present a more detailed 114 view of the influence of both blood components, erythrocytes 115 and plasma.

117
The sedimentation of erythrocytes in blood plasma has been 118 proposed, but was never established, to obey a so-called gel 119 sedimentation regime 11, 12 . Gel sedimentation is efficiently 120 represented by several equations established for colloidal sys-121 tems 13-16 . In particular, gel sedimentation can exhibit a de-122 layed collapse 17, 18 . Here, we apply the gel sedimentation 123 model for sedimenting erythrocytes. This means that erythro-124 cytes first experience only little sedimentation. Then, after 125 a characteristic time on the order of a few minutes, some 126 "cracks" or "channels" form within the aggregated structure 127 of the erythrocytes, as suggested by a previous study 11 . We 128 also observed this behavior in our samples (see Supplemen-129 tal Fig. S5). It is worthwhile to notice that, due to volume 130 conservation, the sedimentation of the erythocytes implies 131 an upward flow of the plasma. The channel structures then 132 enhance this upward flow of liquid. Here, we will apply for 133   Table 1), where 1 patient was measured twice (independent blood sampling on different days). Since none of the controls presented with acanthocytes, they are summarized in one data point with their average ESR. The plotted line represents the linear regression. The analysis reveals a significant correlation for both parameters (panels C and D, R 2 coefficient of 0.61 and 0.59, respectively).

4/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. .

5/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. .
where k 0 is the permeability of the aggregated erythrocyte 138 structure, ∆ρ is the density difference between the erythro-139 cytes and the suspending liquid, φ 0 is the initial hematocrit, To estimate the hole size, we allowed the erythrocytes to and Methods for simulation details). Figure 3I shows the  Fig. S3C) 207 was slightly increased in the NAS patients but, as previously 208 mentioned, would then increase the ESR and therefore cannot 209 explain the differences. Thus, different cases can be compared 210 by scaling the observed maximum sedimentation speed with 211 the hole size; see Eq. 2. If we denote patients with a subscript 212 NAS and the healthy controls with C, then we obtain The v M of control (v C ) and NAS blood (v NAS ) as well as the 214 corresponding erythrocyte-dextran mixtures are shown in Fig. 215 4D (gray and orange columns, respectively). Additionally, 216 the sedimentation speed was scaled according to Eq. 3. This 217 scaling is plotted in Fig. 4D (green columns), along with 218 the error bars corresponding to the standard deviation of the 219 scaled speed measurements. The scaling makes the velocities 220 comparable and indicates that the difference in the aggregate 221 geometry as characterized by hole area accounts for the ob-222 served difference in sedimentation velocity when erythrocytes 223 are suspended in a dextran-based medium. However, the scal-224 ing is not sufficient to explain the velocity ratio when the 225 sedimentation is measured in autologous plasma, as shown in 226 Fig. 1. This indicates that the aggregate geometry accounts 227 for some fraction of the ESR slowdown, while the remain-228 ing fraction must correspond to other contributions, such as 229 patient plasma composition.

6/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020.

7/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. The bars indicate the values of the characteristic erythrocyte sedimentation velocities. For each suspension medium, the patient scaling was obtained by multiplying the characteristic speed of each patient's erythrocytes by the ratio of the characteristic hole areas, as justified by Eq. 3. While the scaled velocity in autologous plasma is significantly smaller than the characteristic speed of the control erythrocytes, the scaling lies within the same range when the autologous plasma is replaced by a dextran-based medium. n.s., not significant (p>0.05); * p<0.05.)

8/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

231
As mentioned above, the total protein content likely con-232 tributes to the reduced ESR for the NAS patients (Supple-233 mental Fig. S4A). Therefore, we further analyzed plasma 234 constituents, such as fibrinogen (Fig. 5A) parameter that could easily complement the error-prone acan-285 thocyte count. Mutations can still be confirmed by molecular 286 biology methods to differentiate among NASs (cp. Materials 287 and Methods). Since the ESR also depends on the geometry 288 of the tubes 26, 27 , systems other than those used here may need 289 their own standardization. 290 Conventionally, ESR is used to detect general inflammation, 291 which is indicated by an increased ESR 28, 29 ; however, there 292 is no conflicting overlap because the change is in the opposite 293 direction in NAS. We would have liked to test whether the 294 ESR correlates to the severity of the disease, but there is no 295 validated index or score for NAS severity available. Similar 296 to the acanthocyte count, ESR cannot discriminate between 297 ChAc and MLS. Furthermore, we cannot make a statement 298 about other NASs (such as pantothenate kinase-associated 299 neurodegeneration (PKAN)), but we suppose a prolongation 300 of the ESR also occurs in blood samples of patients with other 301 diseases showing an increased number of acanthocytes.

302
The number of patients in our study may seem low (6 303 ChAc and 3 MLS patients), but given the ultrararity of these 304 diseases-the estimated prevalence is less than 1 to 5 per 10 6 305 inhabitants each 5 -the number of participants can well be 306 regarded as acceptable. In particular, this holds true because 307 with such a low number of patients, we were able to obtain sig-308 nificant results. Recent studies have identified promising drug 309 targets, such as Lyn kinase, as a potential disease-modifying 310 therapy for ChAc 5 . Considering the rarity and high variability 311 of the natural history of the disease, there is an urgent need for 312 a robust biomarker to achieve clinical trial readiness. The ESR 313 might represent an ideal biomarker candidate in this context. 314 We have shown that both erythrocytes properties and blood 315 plasma composition of NAS patients influence the ESR. Both 316 components reduce the ESR, i.e. they are additive and provide 317 a clear cut-off to differentiate from healthy individuals. This 318 is to some extent surprising since the ESR is a measure that 319 integrates numerous parameters, and its initial intended use as 320 a parameter for reporting inflammation has lost its importance 321 to a certain degree within the last decades 30 due to a lack of 322 specificity.

323
Furthermore, we discussed physical mechanisms related to 324 the erythrocyte and plasma properties that determine the ESR. 325 Historical reports describe an altered aggregation behavior of 326 acanthocytes 31 and the influence of erythrocyte shape on the 327 ESR 32 . Here, we applied the principles of colloidal physics 328 to describe sedimenting erythrocytes in a quantitative man-329 ner. Thus, we clearly show that the geometric properties of 330 erythrocyte aggregates influence the porosity of sedimenting 331 aggregates, which has a direct impact on the sedimentation 332 velocity. Furthermore, a modeling approach provides evi-333 dence that the shapes of the erythrocytes (acanthocytes and 334 echinocytes) and their rigidity are the primary cause for the 335 differences in aggregate geometry. Regarding the plasma 336 composition, we highlighted a difference in fibrinogen con-337 centration, which modifies the interaction energy between 338 erythrocytes. While the detailed influence of this energy on 339 The optical force holding the cells is stepwise decreased, and the overlap distance tends to increase in the same manner. Finally, spontaneous aggregation overcomes the optical forces, and the erythrocytes escape the trap. The trapping force at which the cells aggregate is considered to be the aggregation force. D: Comparison of aggregation force. A significant difference between patients and controls was observed. n.s., not significant (p>0.05); ** p<0.01; *** p<0.001; and **** p<0.0001.

10/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. . the sedimentation process is still under active discussion, it is 340 clearly a key feature in the sedimentation process.

394
Blood smears were prepared as described by Storch et al. 395 2005 10 . Whole EDTA-blood was suspended in a 1:1 ratio in 396 a 0.9 % sodium chloride solution containing 5 IU/ml heparin 397 and incubated for 1 h at room temperature. Then, a drop of 398 this solution was smeared onto a glass coverslip by using a 399 second coverslip. Dried smears were stained according to 400 the Pappenheim method. First, they were placed in May-401 Grünwald solution for 3-4 min and then washed with distilled 402 water. This was followed by staining with Giemsa solution 403 for 20 min followed by another washing step with distilled 404 water. Images were visualized with an Olympus microscope 405 (BX60, Olympus, Japan) with a 40× Plan Fluor Objective 406 and 2× postmagnification and recorded with a CCD camera 407 (DP73, Olympus, Japan). At least 400 cells were counted by 408 visual inspection and classified into stomatocytes, discocytes, 409 echinocytes, acanthocytes and 'other' cell shapes.

411
Approximately 5 μl of blood was placed into 1 ml of 0.1 % 412 glutaraldehyde (Sigma-Aldrich, USA) solution in PBS to fix 413 cells in the shape in which they occur in the circulation 36 . 414 Erythrocytes (5 μl in 1 ml PBS) were stained with 5 μl of 415 CellMaskTM Deep Red plasma membrane stain (0.5 mg/ml; 416 Thermo Fisher Scientific, USA) for 24 h at room tempera-417 ture. Then, the cells were washed 3 times by centrifugation 418 at 4000 rcf for 5 min (Eppendorf Micro Centrifuge 5415 C, 419 Brinkmann Instruments, USA) in 1 ml of PBS solution. Af-420 ter washing, the cells were resuspended in PBS and finally 421 placed on a glass slide for confocal microscopy. Each la-422 beled sample was placed between two glass slides for imaging 423 (VWR rectangular coverglass, 24 × 60 mm 2 ) by employing 424 a piezo stepper for a 20 μm z-range. Confocal image genera-425 tion was performed with a spinning disk-based confocal head 426 (CSU-W1, Yokogawa Electric Corporation, Japan). Image 427 sequences were acquired with a digital camera (Orca-Flash 428 4.0, Hamamatsu Photonics, Japan). A custom written MAT-429 LAB routine was used to crop single cells from each image 430 and perform 3D reconstruction to enable visualization of the 431 3D shape of the cell. Each single-cell 3D image contained 68 432 individual planes with an extent of 100 by 100 pixels and a 433 lateral (x/y) resolution of 0.11 μm/pixel. The piezo stepper 434 had a minimal step width of 0.3 μm, defining the z-resolution. 435 To compensate for the difference in resolution in the x/y and 436 z-directions, we modified the z-scale by means of linear in-437 terpolation. Thus, the obtained z-stack had dimensions of 438 100 × 100 × 185 voxels. The image stacks were then passed 439 to a custom written ImageJ script. By applying a fixed thresh-440 old for every image, the script binarized the confocal z-stack 441 to retrieve the cell membrane as an isosurface.

11/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. .   the potential, ε, was selected from the range of 1.5 − 2.5 k B T 497 for different simulations to reflect aggregation variability in 498 experiments. The total hematocrit φ was set to 50 %. Two 499 sets of simulations were performed, including one with only 500 healthy erythrocytes and one with a mixture of 80 % healthy 501 cells and 20 % acanthocytes (NAS case) by area fraction. Each 502 set consisted of 11 simulations with different parameter val-503 ues. Note that since the area of an acanthocyte is larger than 504 that of a healthy cell, the total number of cells in the NAS 505 case is smaller than that in the healthy case.

507
The aggregation forces between erythrocytes were mea-508 sured through holographic optical tweezers as previously de-509 scribed 39 . This is a versatile technique for measuring forces 510 in the piconewton (pN) range and allows force measurement 511 at a single-cell level. A single infrared beam Nd:YAG laser 512 (1064 nm, 3 W, Ventus 1064, Laser Quantum, UK) is reflected 513 by a parallel aligned nematic liquid crystal spatial light mod-514 ulator (PAL-SLM, PPM X8267-15, Hamamatsu Photonics, 515 Japan) and focused with a large numerical aperture oil immer-516 sion objective (60×, Nikon, Japan) in an inverted microscope 517 (Nikon, TE 2000, Japan). The real-time manipulation of op-518 tical traps was made possible by imaging the sample with a 519 CMOS camera (ORCA Flash 4.0 V3, Hamamatsu, Japan) and 520 using a MATLAB routine to compute the phase. The optical 521 force holding cells can be tweaked by varying the initial laser 522 power. Measurements were carried out following the protocol 523 depicted in Fig. 5C. For each measurement, two erythrocytes 524 with a discocytic shape were manually selected. Each erythro-525 cyte was held by two optical traps placed at their extremity. 526 The lower erythrocyte was brought into contact with the up-527 per erythrocyte by changing the vertical position of the traps 528 holding the erythrocyte in steps of 1 μm/s. The overlapping 529 contact length between the erythrocytes was initially set to 530 4.5 μm for all the measurements. The aggregation force corre-531 sponds to the force that is insufficient to counterbalance the 532 spontaneous aggregation forces between the two erythrocytes. 533 The same protocol was repeated to measure at least 7 pairs 534 of cells in both the pathological and healthy samples. Optical 535 tweezers were calibrated, and forces were determined based 536 on Stokes' law as in 40 .

538
Statistical analysis was performed in Prism8 (GraphPad Soft-539 ware, USA). All data sets were checked for normality of the 540 distribution by the Shapiro-Wilk test. The ESR for controls, 541 ChAc and MLS patients was evaluated by the Brown-Forsythe 542 and Welch ANOVA tests. Further significance between two 543 conditions was then performed by an unpaired t-test, except if 544 otherwise stated in the figure legend. Significance p-values are 545 abbreviated as n.s. (not significant) for p>0.05, * for p<0.05, 546 ** for p<0.01, *** for p<0.001 and **** for p<0.0001. Cor-547 relation analysis was performed by simple linear regression, 548 and the p-value provides the slope difference from zero.  Figure S1. Sex-specific comparisons. A: Sex-specific comparison of the ESR after 2 h. In the control group, males and females are equally distributed, while in the patient group, the higher prevalence of the disease in males is reflected by the presence of only one female ChAc patient. We did not detect any significant sex-based difference between the controls and patients. B: Comparison of hematocrit. Since gender differences in the hematocrit are known, we compared this parameter separately for males and females between control and NAS blood samples. We did not find significant differences between healthy controls and NAS patients either in the gender-specific comparison as outlined here or when male and female subjects were pooled. Figure S2. Analysis of blood smears. The blood smear from patient ChAc-5 was analyzed by two skilled staff members familiar with erythrocyte shapes. They were blinded to the intended comparison when they performed the counts. Approximately 1000 erythrocytes were counted by each analyzer. While the number of discocytes was consistent, the number of echinocytes, acanthocytes and other erythrocyte shapes (including stomatocytes and other cell shapes of known or unknown categories) varied tremendously. These differences reflect the uncertainty of acanthocyte counting based on blood smears.

15/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020.

16/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. Comparison of albumin concentration. For the albumin concentration, there was no significant difference between healthy controls and NAS patients. C: Comparison of the C-reactive protein concentration. For the C-reactive protein (CRP) concentration, there was no significant difference between healthy controls and NAS patients. D: Comparison of the immunoglobulin G concentration. For the immunoglobulin G (IgG) concentration, there is no significant difference between healthy controls and NAS patients.

17/18
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint this version posted September 3, 2020. . Figure S5. Channels that appear in sedimenting blood. The cell glass is approximately 100 μm thick, which makes the cracks visible to the naked eye. The square panel is a magnified view of the squared area in the rectangular picture. These channels are characteristic of the transient gel sedimentation regime of colloidal suspensions.   Figure S7. Comparison of fresh and delayed observations. A: Comparison of ESR time traces -fresh blood vs. 6 h after withdrawal. Blood samples from one healthy control and patient ChAc-5 were measured immediately after blood withdrawal and 6 h later. The delayed curves are almost indistinguishable from the fresh measurements.B: Comparison of the ESR after 2 h -fresh measurements vs. approximately 6 h of transportation. A subpopulation of the patients traveled to our laboratory in Saarbrücken to allow blood draws for immediate measurement. For the other patients, blood was collected at their local hospitals. We detected no difference in the ESR between fresh and transported blood for either the healthy controls or the NAS patients.