top of page

Young Plasma Infusions Significantly Improve Clinical Symptoms and UPDRS Scores in Patients with Parkinson’s Disease

by Dian J Ginsberg, MDa and Carl Scheffey

a) The Ginstitute of Functional Medicine, 6750 West Loop South, Suite 425, Bellaire, Texas 77401. dianginsberg.md@gmail.com

b) Kennard Institute, 2308 Mt Vernon Ave, STE 426, Alexandria, Virginia 22301

Abstract

In both preclinical and clinical studies, transfusions of plasma from young individuals have been reported to ameliorate aspects of neurodegeneration. This study was designed as a preliminary test of the hypothesis that plasma transfusions from young donors might benefit Parkinson's patients. 19 patients were allocated to receive either 2-liters of plasma from young donors, in two doses spaced two days apart, or two doses of placebo. For the next 24 weeks, this double-blind study evaluated changes on a modified MDS-UPDRS scale, along with blood tests and other observations. Adverse events possibly related to transfusion were mild rise in blood pressure and urticaria. A t-test on the changes in the sum of UPDRS subscales 1-3 showed that the plasma patients did better than the placebo patients (p = 0.03*). For patients given yFFP® (young Fresh Frozen Plasma), the estimated decrease in the sum of scales 1-3 was 7.1 (95% conf. interval 4.3 to 9.9). Our results give a preliminary indication that young plasma transfusions reduce Parkinson's symptoms and have a place in treatment of these patients. (NCT 04202757).

 

Key Words:

Young Plasma

Parkinson’s Disease

UPDRS Scale

Neurodegenerative

Disease Transfusion

 

1. Introduction

Medicine has been attempting to find a cure for Parkinson’s Disease (PD) since James Parkinson described the ‘Shaking Palsy’ in 1817[1]. Treatment to this date has been targeted medication ‘at’ the neuronal dysfunction that enables the patient to function as best as possible while the disease progresses. Gene therapy is difficult as many limitations hamper the successful development of 2 these candidates, including the need for difficult surgical guided placement, as well as the lack of biomarkers available for diagnosis and therapy management[2,3].

 

A common feature of PD and dementias is chronic neuroinflammation, mitochondrial dysfunction and oxidative stress[4–6]. Neuromelanin synthesis is a protective antioxidant mechanism in the brain to remove potentially toxic oxidized byproducts in the substantia nigra. The protective mechanism can be overburdened as toxic load becomes unsustainable for the cell. Alpha-synuclein and Lewy bodies build up, destroying its proteostasis, mitochondrial function and overall ability to survive. Eventually enough of the dopaminergic neurons in the substantia nigra will die and PD symptoms will develop.

 

Preclinical and clinical research have recently accumulated evidence that the plasma of younger individuals appears to contain rejuvenating properties that can have an impact on degenerative disease. In the present paper, we refer to the plasma of younger individuals as "Young Plasma." In clinical studies, we use the acronym yFFP (young Fresh Frozen Plasma). The evidence raises the question: can neurodegenerative diseases be reversed? Molecular pathways lost in neurodegenerative disease might be reinstated by "the transfer of youthful factors into relevant neural niches” [7]. In the mouse hippocampus, Young Plasma has been reported to reverse a variety of age-related changes[8]. Similar improvements have been seen in other rodent trials [9– 13]. Clinical human studies have reported related findings with yFFP or plasma exchange[14–17].

 

The safety of transfusions with plasma from young donors is already established. The American Red Cross estimates 10,000 units of plasma are used daily in the United States[18]. An estimated 19.8% of blood donations came from the age group 16-24 years, and transfusions have had an excellent safety record[19,20].

 

The primary objective of our study was to assess effectiveness of infusions of two liters of yFFP on Parkinson’s patients. The primary objective of our study was to assess effectiveness of infusions of two liters of yFFP on Parkinson’s patients. The exploratory objectives were clinical improvements using: UPDRS, the Stanford Presenteeism Score (SPS-6)[21], and laboratory parameters.

 

2. Methods

2.1. Enrollment and treatment allocation

 

The target enrollment for the study was 20 patients. Inclusion criteria were: mild to moderate Parkinson’s Disease (less than 3 on Hoehn-Yahr scale), with the ability to do a self-evaluation or have a caregiver to perform the evaluation. Patients with certain severe comorbidities were excluded. The study's protocol and informed consent form were evaluated and approved by the Institute of Cellular and Regenerative Medicine (IRCM) in Santa Monica, Ca. IRB organization number IORG0007913. Patients to be enrolled had all their questions answered and signed an informed consent to participate in the study.

 

The treatment under study was two transfusions of yFFP given on each of two days, with one intervening day between the first and second transfusions. The volume of each transfusion was 3 12.5 ml-per-kg-1 . Donors were matched with patients by blood type and sex. Plasma was slowly infused in order to minimize transfusion-associated circulatory overload (TACO). Patients allocated to placebo received the same volume of fluid on the same schedule, but the fluid was isotonic NaCl with 0.1% pyridoxine (vitamin B6). This enabled the fluid to be a yellow color to even further ensure that even with the mask bag cover the patient would be blinded.

 

The patients and all persons who had direct contact with the patients were blinded as to the treatment allocation. That includes the neurologists who did the outcome measure evaluations.

 

When possible, alternating consecutive patients were allocated to receive yFFP or placebo. As yFFP can be in short supply and blood type matching is critical, it was understood that allocation to yFFP might depend upon the blood type available.

 

2.2. Plasma Collection

 

FDA approved [21CFR640.30] Plasma was collected from sex-identified individuals 18-25 years of age (yFFP®), by Spectrum Plasma. Apheresis machines were used for all collections, which were immediately processed and then held frozen at -30°C. In addition to the twelve standard plasma tests required by FDA regulation (21CFR610.40), Spectrum Plasma’s standard operating procedure is to perform five more tests to ensure safety: all candidates are pretested before becoming eligible donors; mean corpuscular volume, hemoglobin and hematocrit are tested; a serum protein electrophoresis test is performed before a donation is made to confirm the donor has acceptable total protein levels; all plasma collected from female donors is tested for the presence of HLA antibodies; to avoid latent infections, yFFP is held for a minimum of thirty days, within which the donor is required to make another donation that passed all tests, before the first donation is released. All shipments are frozen, temperature monitored and then warmed at the time and point of use.

 

2.3. Data Collection

 

The explored outcomes were documented at baseline then at 4, 12, and 24 weeks. Evaluations were done via a modified form of MDS-UPDRS, the Stanford Presenteeism Scale (SPS-6), and laboratory parameters. Additionally, each patient was also given a blank sheet of paper and the option to journal their overall feelings and any changes they personally wanted to share.

 

The UPDRS scale used was modified for subscale 3. The modified form used in the UPDRS subscale 3 recorded only one score for certain groups of questions. For example, only one score was recorded for the first five questions of the subscale, all of which are on the subject of resting tremor. The score recorded was the maximum for the answers to the five questions. For bradykinesia questions with two bilateral scores, only the maximum of the two scores was recorded. Accordingly, the maximum for subscale 3 was 56. The minimum clinically important difference for the UPDRS scale used was not the same as for the full MDS-UPDRS.

 

4 2.4. Statistical Analysis

 

In advance of computing any statistics, the authors agreed that any p-value less than 0.1 (with twosided alternative) would be considered a preliminary result worth further exploration. We reserve the phrase "statistically significant" to mean p < 0.05.

 

Statistics in this study were based on comparing the outcome measures collected at baseline with those at 24 weeks. The change in scores was evaluated for the two groups of patients with each treatment. Additionally, the populations of changes in the two groups were compared with each other. For both of those kinds of comparisons, the null hypothesis was of zero difference between two group means. For each comparison, we report two-sided p-values for a permutation test[21]. Monte Carlo computations used 108 permutations. For each permutation, the test statistic was the difference between means in the two subpopulations of the pooled data. We also report results of t-tests (with the equal variance assumption), which are parametric, along with parametric estimates for the differences in means between the compared groups. The t-test is familiar to more readers, and affords comparison between the present study and the many studies in the literature that have used t-tests to interpret changes in the UPDRS and SPS scales. Strictly speaking, the assumptions of the t-test are not satisfied for these discrete data (see Fig. 2). It is, nevertheless, a useful tool. The permutation test is nonparametric, and makes weaker assumptions about the origins of the data. Each population was checked for data characteristics such as extreme skewness that render permutation tests inappropriate[22].

 

Statistical computations used the R statistical language[23], with the "coin" add-on package[24].

 

3. Results

 

3.1 Enrollment

 

Fig. 1 shows a CONSORT-style patient flow diagram. Of the 22 patients enrolled, three withdrew for personal reasons prior to study initiation. Nine patients were allocated to receive yFFP, and ten patients were allocated to placebo.

Fig1 Recruitment Letter Sent To Patients

3.2. Baseline Patient Characteristics

 

Tables 1 and 2 compare the treatment groups with respect to certain demographics and baseline patient characteristics. Most of the patient samples were males over the age of 60 years. There were not dramatic differences seen between the allocation groups.

Table 1. Baseline patient characteristics (part 1).

Table 1 Baseline Patient characteristics

Table 2. Baseline patient characteristics (part 2).

Table 2 Baseline Patient Characteristics

3.3 Changes on individual scales

 

For each scale, the patient scores were compared between baseline and 24 weeks. The results are shown in Table 3. For patients treated with yFFP, there were statistically significant decreases in UPDRS subscales 1 and 2.

Table 3. Statistics comparing baseline values with 24-week values, within each treatment group.

Footnote: a

*The one-sided test has p < 0.05. See comment in section 4.

 
Table 3 Statistics comparing 24 week values

3.4. Comparing UPDRS and SPS Changes

 

Table 4 shows a more direct comparison of yFFP with placebo. For each patient, the score on that scale at baseline was subtracted from the score at 24 weeks. That difference was the change in the measure during the study. The population of those differences for yFFP patients was statistically compared with the population of those differences for placebo patients. The comparison was statistically significant for the sum of UPDRS subscales 1-3, and for the SPS scale.

Table 4. Comparing treatment groups for size of change from baseline to 24 weeks.

Footnote: a

*The one-sided test has p < 0.05. See comment in section 4.

 
Table 4. Comparing treatment groups for size of change from baseline to 24 weeks

Figures 2 and 3 show histograms of the some of the differences that were used for the statistical tests of Table 4. Figure 2 shows the differences for the sum of all three of the UPDRS subscales 1-3. Fig. 3 shows the differences for the SPS-6 scale. In each pair of histograms, the top half of the figure shows the distribution of placebo changes from baseline to 24 weeks, and the bottom 8 histogram shows the distribution of yFFP changes. In Fig. 2, more of the yFFP patients showed improvement (negative numbers) for the UPDRS at 24 weeks, minus baseline. Fig. 3 shows a similar comparison for the changes in SPS-6 scores.

Fig. 2. Histograms of changes in sum of UPDRS subscales.

 
Fig. 2. Histograms of changes in sum of UPDRS subscales

Fig. 3. SPS-6 score change histograms.

Placebo and yFFP changes

3.5. Safety

 

There were 5 total mild adverse events in the plasma group and none in the placebo group. Two of the patients, one with hives and one with a rise in blood pressure, chose with the nurse not to complete the infusions and subsequently withdrew from the study. Of the other 3 patients, one patient who was already on hypertensive medications had a small rise in blood pressure (baseline was 132/77 mm Hg (17.6/10.3 kPa) and max was 165/87 mm Hg (22.0/11.6 kPa). The blood pressure normalized before the completion of the infusion. Two patients experienced some itching but neither showed signs of urticaria and all symptoms resolved with IV diphenhydramine and oral fexofenadine. None of the patients had any deleterious changes in their laboratory values.

4. Discussion

This study presents evidence that the administration of young Fresh Frozen Plasma significantly improves multiple outcome measures during the 24 weeks after administration. The comparisons of Table 4 show statistically significant results that patients assigned to yFFP improved more than patients assigned to placebo, with respect to the sum of UPDRS subscales 1-3 and also SPS-6.

 

The most common risks associated with FFP include: transfusion related acute lung injury, circulatory overload, anaphylactic reactions, transmission of infections and hemolytic transfusion reactions[25]. Because precautions were taken to avoid those risks, we believe that most appropriate statistical analysis would assume a priori that yFFP could not worsen the outcome measures, in the general pool of patients used for enrollment. Under that assumption, one-sided rather than two-sided tests would be appropriate and p-values would be cut in half. We note in tables 3 and 4 the tests that then become statistically significant in the sense of p < 0.05.

 

As the dopamine loss within the central nervous system creates lack of both muscle use and imbalanced forces over time, peripheral dysfunction in the muscle will occur. Therefore, the motor changes may require more time and muscle rehabilitation to show a significant improvement[26,27]. We hypothesize that the plasma crossing the blood brain barrier partially reversed the inflammatory process of the neurodegenerative disease. It may be that yFFP replaces as well as modulates production of these factors in the brain such as brain derived neurotrophic factor, antioxidant support, oxytocin, thrombospondin-4 and SPARC-like protein 1 as well as upregulating enzymes such as enzyme ten eleven translocation methylcytosine dioxygenase 2 (Tet2), known to support neurogenesis[28–31]. Many of these components have a very short halflife so consideration that supplying smaller aliquots of yFFP weekly perhaps will be a better future option. As apheresis and the removal of senescent and inflammatory cells has been reported to give very promising results[15,32], combining this removal process with the addition of young factors present only in yFFP may prove to be optimal therapy to address both the improvement of disease states as well as provide a mechanism for healthy aging. We refer to this combination approach as ‘Rectangularization of Aging.’

 

The young Fresh Frozen Plasma (yFFP) used in this study was collected by methods intended to safeguard the factors described in the above paragraph. In general, typical blood bank plasma is collected from donors of all ages without sex-segmentation, is often kept for up to eight hours as whole blood and in some collections, treated with methylene blue and UV light. It has been shown that those legal, but inferior methods of collection can incur as much as a 58% reduction in important factors and proteins[33,34]. It has also been shown that there are significant differences in the plasma proteome between sexes and ages[35].

 

Laboratory parameters did not show that the patients had any harmful effects from the plasma. While our study did have 2 patients show a mild reaction requiring discontinuation of the infusion, no severe reactions were recorded.

 

This preliminary study has limitations. The sample was small. The plasma/placebo treatment allocation was largely done by alternation between consecutive patients, rather than randomization. UPDRS evaluations were not done at a consistent time of day and were scheduled without regard for the medication schedules of the individuals. Patients were recruited via an email sent from two large neurology practices, and the volunteers who qualified became the sample.

In spite of the study limitations, the results support the hypothesis that yFFP has benefit to Parkinson’s patients. Further studies with larger groups of patients and comparing therapies need to be explored.

Acknowledgements

Patient evaluations were performed by the neurologists: Igor Cherches, MD and Eddie Patton Jr, MD. This work was supported by a grant from the Carolina Longevity Institute. Spectrum Plasma donated the yFFP and supported preparation of the manuscript. The sponsors and participating physicians all contributed to the study design. No sponsor analyzed or contributed to the interpretation of any data.

 

Declaration of Interest

 

Dian J. Ginsberg, MD is a shareholder in Spectrum Plasma, Inc. Carl Scheffey, PhD received consulting fees from Spectrum Plasma, Inc. The acronym yFFP is a registered trademark of Spectrum Plasma, Inc. Authors contributions Dian J. Ginsberg, MD: Supervision, Funding acquisition, Project administration, Conceptualization, Methodology, Investigation, Data curation, Visualization, Writing-original and review. Carl Scheffey: Formal analysis, Software, Methodology, Data curation, Visualization, Writing-- original and review.

 

References

[1] Goetz CG. The history of Parkinson’s disease: Early clinical descriptions and neurological therapies. Cold Spring Harb Perspect Med 2011;1. https://doi.org/10.1101/cshperspect.a008862.

[2] Colombo D, Pnevmatikou P, Melloni E, Keywood C. Therapeutic innovation in Parkinson’s disease: a 2020 update on disease-modifying approaches. Expert Rev Neurother 2020;20:1047–64. https://doi.org/10.1080/14737175.2020.1800454.

[3] Bartus RT, Weinberg MS, Samulski RJ. Parkinson’s disease gene therapy: Success by design meets failure by efficacy. Mol Ther 2014;22:487–97. https://doi.org/10.1038/mt.2013.281.

[4] Lin MT, Beal MF. Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases. Nature 2006;443:787–95. https://doi.org/10.1038/nature05292.

[5] Balestrino R, Schapira AHV. Parkinson disease. Eur J Neurol 2020;27:27–42. https://doi.org/10.1111/ene.14108.

[6] Harms AS, Kordower JH, Sette A, Lindestam Arlehamn CS, Sulzer D, Mach RH. Inflammation in Experimental Models of α-Synucleinopathies. Mov Disord 2021;36:37–49. https://doi.org/10.1002/mds.28264.

[7] Castellano JM, Kirby ED, Wyss-Coray T. Blood-borne revitalization of the aged brain Aging as a multi-system problem. JAMA Neurol 2015;72:1191–4. https://doi.org/10.1001/jamaneurol.2015.1616.Blood-borne.

[8] Li N, Zhang X, Gu Z, Su C, Lian H. Young plasma attenuates cognitive impairment and the cortical hemorrhage area in cerebral amyloid angiopathy model mice. Ann Transl Med 2021;9:147–147. https://doi.org/10.21037/atm-20-8008.

[9] Zhao Y, Qian R, Zhang J, Liu F, Iqbal K, Dai CL, et al. Young blood plasma reduces Alzheimer’s disease-like brain pathologies and ameliorates cognitive impairment in 3×Tg-AD mice. Alzheimer’s Res Ther 2020;12:1–13. https://doi.org/10.1186/s13195-020-00639-w.

[10] Katsimpardi L, Litterman NK, Schein PA, Miller CM, Loffredo FS, Wojtkiewicz GR, et al. Vascular and Neurogenic Rejuvenation of the Aging Mouse Brain by Young Systemic Factors. Science (80- ) 2014;344:630–4. https://doi.org/10.1126/science.1251141.

[11] Rebo J, Mehdipour M, Gathwala R, Causey K, Liu Y, Conboy MJ, et al. A single heterochronic blood exchange reveals rapid inhibition of multiple tissues by old blood. Nat Commun 2016;7:1–11. https://doi.org/10.1038/ncomms13363.

[12] Yuan JJ, Zhang Q, Gong CX, Wang FX, Huang JC, Yang GQ, et al. Young plasma ameliorates aging-related acute brain injury after intracerebral hemorrhage. Biosci Rep 2019;39:1-12. https://doi.org/10.1042/BSR20190537.

[13] Katsimpardi L, Rubin LL. Young systemic factors as a medicine for age-related neurodegenerative diseases. Neurogenesis 2015;2:e1004971. https://doi.org/10.1080/23262133.2015.1004971.

[14] Costa M, Páez A. Emerging insights into the role of albumin with plasma exchange in Alzheimer’s disease management. Transfus Apher Sci 2021;60. https://doi.org/10.1016/j.transci.2021.103164.

[15] Mehdipour M, Etienne J, Liu C, Mehdipour T, Kato C, Conboy M, et al. Attenuation of age-elevated blood factors by repositioning plasmapheresis: A novel perspective and approach. Transfus Apher Sci 2021;60:103162. https://doi.org/10.1016/j.transci.2021.103162.

[16] Parker JE, Martinez A, Deutsch GK, Prabhakar V, Lising M, Kapphahn KI, et al. Safety of Plasma Infusions in Parkinson’s Disease. Mov Disord 2020;35:1905–13. https://doi.org/10.1002/mds.28198.

[17] Sha SJ, Deutsch GK, Tian L, Richardson K, Coburn M, Gaudioso JL, et al. Safety, Tolerability, and Feasibility of Young Plasma Infusion in the Plasma for Alzheimer Symptom Amelioration Study: A Randomized Clinical Trial. JAMA Neurol 2019;76:35–40. https://doi.org/10.1001/jamaneurol.2018.3288.

[18] US Blood Supply Facts n.d. https://www.redcrossblood.org/donate-blood/how-to-donate/how-blood-donations-help/blood-needs-blood-supply.html#:~:text=Facts About Blood Needs&text=Approximately 36%2C000 units of red,needed daily in the U.S.&text=Nearly 21 million blood components are t (accessed November 1, 2021).

[19] Mowla SJ, Sapiano MRP, Jones JM, Berger JJ, Basavaraju S V. Supplemental findings of the 2019 National Blood Collection and Utilization Survey. Transfusion 2021;61. https://doi.org/10.1111/trf.16606.

[20] Savinkina AA, Haass KA, Sapiano MRP, Henry RA, Berger JJ, Basavaraju S V., et al. Transfusion-associated adverse events and implementation of blood safety measures - findings from the 2017 National Blood Collection and Utilization Survey. Transfusion 2020;60:S10–6. https://doi.org/10.1111/trf.15654.

[21] Koopman C, Pelletier KR, Murray JF, Sharda CE, Berger ML, Turpin RS, et al. Stanford Presenteeism Scale: Health Status and Employee Productivity. J Occup Environ Med 2002;44:14–20. https://doi.org/10.1097/00043764-200201000-00004.

[22] Manly BFJ. Randomization, bootstrap, and Monte Carlo methods in biology. Second. Chapman & Hall; 1997.

[23] Christensen WF, Zabriskie BN. When Your Permutation Test is Doomed to Fail. Am Stat 2021;0:1–28. https://doi.org/10.1080/00031305.2021.1902856.

[24] Team RC. R: A Language and environment for statistical computing. Vienna, Austria: 2020.

[25] Hothorn T, Hornik K, Wiel MA van de, Zeileis A. Implementing a Class of Permutation Tests: The coin Package. J Stat Softw 2008;28. https://doi.org/10.18637/jss.v028.i08.

[26] Pandey S, Vyas GN. Adverse effects of plasma transfusion. Transfusion 2012;52:65S-79S. https://doi.org/10.1111/j.1537-2995.2012.03663.x.

[27] Cano-De-La-Cuerda R, Pérez-De-Heredia M, Miangolarra-Page JC, Muñoz-Hellín E, Fernández-De-Las-Peñas C. Is there muscular weakness in parkinson’s disease? Am J Phys Med Rehabil 2010;89:70–6. https://doi.org/10.1097/PHM.0b013e3181a9ed9b.

[28] Inkster LM, Eng JJ, MacIntyre DL, Jon Stoessl A. Leg muscle strength is reduced in Parkinson’s disease and relates to the ability to rise from a chair. Mov Disord 2003;18:157–62. https://doi.org/10.1002/mds.10299.

[29] Gontier G, Iyer M, Shea JM, Bieri G, Wheatley EG, Ramalho-Santos M, et al. Tet2 Rescues Age-Related Regenerative Decline and Enhances Cognitive Function in the Adult Mouse Brain. Cell Rep 2018;22:1974–81. https://doi.org/10.1016/j.celrep.2018.02.001.

[30] Xu R, Gong CX, Duan CM, Huang JC, Yang GQ, Yuan JJ, et al. Age-Dependent Changes in the Plasma Proteome of Healthy Adults. J Nutr Health Aging 2020;24:846–56. https://doi.org/10.1007/s12603-020-1392-6.

[31] Lommatzsch M, Zingler D, Schuhbaeck K, Schloetcke K, Zingler C, Schuff-Werner P, et al. The impact of age, weight and gender on BDNF levels in human platelets and plasma. Neurobiol Aging 2005;26:115–23. https://doi.org/10.1016/j.neurobiolaging.2004.03.002.

[32] Mehdipour M, Etienne J, Chen C-C, Gathwala R, Rehman M, Kato C, et al. Rejuvenation of brain, liver and muscle by simultaneous pharmacological modulation of two signaling determinants, that change in opposite directions with age. Aging (Albany NY) 2019;11:5628–45. https://doi.org/10.18632/aging.102148.

[33] Johnson AA, Shokhirev MN, Wyss-Coray T, Lehallier B. Systematic review and analysis of human proteomics aging studies unveils a novel proteomic aging clock and identifies key processes that change with age. Ageing Res Rev 2020;60:101070. https://doi.org/10.1016/J.ARR.2020.101070.

[34] Dogra M, Sidhu M, Vasudev R, Dogra A. Comparative analysis of activity of coagulation Factors V and VIII and level of fibrinogen in fresh frozen plasma and frozen plasma. Asian J Transfus Sci 2015;9:6. https://doi.org/10.4103/0973-6247.150936.

[35] Lozano M, Cid J, Müller TH. Plasma Treated with Methylene Blue and Light: Clinical Efficacy and Safety Profile. Transfus Med Rev 2013;27:235–40. https://doi.org/10.1016/J.TMRV.2013.08.001.

[36] Lehallier B, Gate D, Schaum N, Nanasi T, Lee SE, Yousef H, et al. Undulating changes in human plasma proteome profiles across the lifespan. Nat Med 2019;25:1843–50. https://doi.org/10.1038/s41591-019-0673-2.

bottom of page