Hemostatic and platelet proteins are associated with bleeding grade and time to major bleed in hemato-oncological patients.

Introduction

Hemato-oncological patients undergoing chemotherapy and stem cell transplantation experience periods of bone marrow aplasia. This leads to several clinical complications including thrombocytopenia, which is commonly associated with the occurrence of bleeding. Although life-threatening bleeds only occur in <5% of cases, major bleeding complications, classified as World Health Organization (WHO) grade ≥2, remain common in hemato-oncological patients (19%-50%) 1 , 2 and have been associated with higher risk of death within the first 60 days after acute myeloid leukemia (AML) diagnosis. 3 To manage bleeding events, prophylactic platelet transfusions are often administered to patients, and numerous clinical trials have focused on optimizing transfusion guidelines to improve the prevention and treatment of bleeding. 1 , 2 , 4 , 5 , 6

Transfusion guidelines in hemato-oncological patients are based on platelet counts. This is supported by reports of more frequent bleeding events in patients receiving therapeutic-only platelet transfusions than in those receiving prophylactic transfusions 7 and increased bleeding risk in severely thrombocytopenic patients (≤5 × 10 9 /L). 8 Nonetheless, the association between platelet counts and bleeding remains open to questions because this relationship is not observed at platelet counts >5 × 10 9 /L nor at lowest daily platelet counts. 8 , 9 Similarly, no differences in severe bleeding events nor in the time to recovery from thrombocytopenia between therapeutic-only platelet transfusions compared to prophylactic transfusions have been observed. 1 , 7 Additionally, other clinical parameters, including disease and fever, do not have a clear association with bleeding outcome. 7 This is in agreement with a post hoc analysis of clinicopathological and demographic characteristics in the Trial of Prophylactic Platelets, which found that baseline risk factors have a low discriminative ability to predict bleeding. 10 Overall, the current screening of clinicopathological characteristics fails to identify hemato-oncological patients at risk of major bleeding. This emphasizes the multifactorial nature of bleeding risk and the need to further dissect the pathophysiology of bleeding complications.

We have recently shown that elevated C-reactive protein plasma levels and albuminuria may be associated with bleeding phenotype, highlighting the potential of additional biomarkers beyond platelet counts. 11 In addition, we have identified systemic dysregulations in patients with AML during treatment using a plasma proteomics approach. 12 This highlights the opportunity to apply state-of-the-art technologies to better understand the pathophysiological changes in hemato-oncological patients, focusing on bleeding events, with the ultimate aim to improve their clinical monitoring. Here, we integrated mass spectrometry (MS)–based plasma profiling with retrospective clinical data obtained from the Pathogen Reduction Evaluation and Predictive Analytical Rating Score (PREPAReS) study to identify molecular signatures associated with bleeding events across hemato-oncological patients.

Methods

Study design and selection of samples

Patients from the control arm, who only received standard platelet concentrate products, of the PREPAReS study (registered at The Netherlands Trial Register identifier: NTR2106; and ClinicalTrilas.gov identifier: NCT02783313 ) were included. The clinical design, inclusion criteria, and ethical approval of the PREPAReS study have been previously described. 13 , 14 The selected patients had different diagnosis, including AML, acute lymphoblastic leukemia (ALL), multiple myeloma (MM), non-Hodgkin lymphoma (NHL), chronic myeloid leukemia (CML), mantle cell lymphoma (MCL), and chronic lymphocytic leukemia (CLL). For a small group of patients, the diagnosis was determined as undefined. These patients received different treatments, including remission induction chemotherapy, consolidation chemotherapy, allogeneic stem cell transplantation, or autologous stem cell transplantation. From these, patients with available plasma samples and retrospective information on disease, treatment, and WHO scored bleeding records were included ( supplemental Figure 1 ; Table 1 ).

MS-based proteomic profiling

Plasma samples from 176 patients and platelet samples from 3 healthy volunteers were analyzed with MS-based proteomic profiling. Briefly, proteins from samples were denatured, reduced, and alkylated and subsequently digested into peptides. Peptides were analyzed using an Evosep One LC system (Evosep, Odense, Denmark) coupled to a timsTOF HT (Bruker Daltonics, Bremen, Germany), operating in data-independent acquisition using parallel accumulation-serial fragmentation. Details on sample preparation, MS-based analysis, and data processing are provided in supplemental Methods .

Statistical analysis

The workflow stability was evaluated with coefficients of variations (CVs) calculated per protein from data of repeated quality control (QC) sample injections, without log 2 transformation (n = 21). Data from clinical records regarding platelet counts, bleeding events, and platelet transfusions were analyzed to define the number of days with thrombocytopenia or major bleeding events (WHO grade ≥2) and the number of platelet transfusions per patient. Data from daily records were compiled per patient, with number of records counted. Kruskal-Wallis rank sum test with the kruskal.test function from the ggpubr R package was used for comparisons between multiple groups. Pairwise comparisons were performed with Dunn test with Benjamini-Hochberg (BH) multiple comparison adjustment with the function dunn_test from the package rstatix. 15 Statistical analyses were performed on plasma samples with information available on bleeding records and bleeding grade. Plasma samples and bleeding records were matched based on date. Specific time information for both records and sample collection was unavailable.

Principal component analysis was performed for all proteins identified in all samples before imputation. Bleeding grade was included as a categorical ordinal outcome with 3 levels: no bleeding, minor bleeding (WHO grade 1), and major bleeding (WHO grade ≥2). 1 , 2 , 16

The association between bleeding grade and protein abundance was investigated using an ordinal logistic regression including age, biological sex, and blood type as covariates for each identified protein using imputed data. These associations were tested assuming proportional odds for all proteins. Estimates of association were obtained as odds ratios (OR) with the corresponding 95% confidence interval (CI). Two-sided P values were computed based on t values, and associations were considered to be statistically significant after multiple testing with Benjamini-Hochberg (<.05). The ordinal logistic regression was performed using the polr function from the MASS R package 17 (version 7.3-54). To assess the robustness of the associations for each protein given the data, bootstrapping with 10 000 iterations was applied to estimate the distribution of the point estimates. Kruskal-Wallis test was performed, as previously mentioned, for all significant proteins from the ordinal logistic regression (n = 55), with Benjamini-Hochberg correction for multiple testing. Significant results from the Kruskal-Wallis test were further analyzed with Dunn test for pairwise comparisons, with the function and package earlier described. Correlation analysis with all significant proteins (n = 55) was performed by calculating Pearson correlation based on imputed values for all 55 proteins with the Hmics package. 18 Protein networks were defined based on a Pearson correlation threshold of 0.6 and visualized in Cytoscape. 19

Moderated t tests were performed using Limma 20 to compare protein levels in first collected samples between patients with or without a recorded major bleed. Proteins were considered to be statistically significant after multiple testing with Benjamini-Hochberg (≤.05) and absolute log fold change ≥1. The association between protein abundance and time to the nearest major bleeding event was investigated using a Cox proportional hazard model including age, biological sex and blood type as covariates for all significant proteins identified by the ordinal logistic regression using imputed data. Hazard ratios (HR) and 95% CI were computed using the coxph function from the survival R package 21 (version 3.7). Time to event was defined as the number of days between first plasma available and the first recorded major bleeding event or last follow-up, whichever occurred first. Considering the timing of bone marrow aplasia and the associated risk of severe bleeding in hemato-oncological patients, time to event was censored at 30 days after randomization. Patients (n = 16) with a first major bleeding event before the first plasma collection were excluded from this analysis. Kaplan-Meier curves for specific proteins separating patients with high and low abundance were obtained by dividing samples above or below the median log 2 label-free quantification values for each protein across all samples. Kaplan-Meier curves of time to major bleed between groups of patients who did or did not receive a platelet transfusion before first sample collection were obtained. Spearman correlations between imputed values for all proteins with platelet counts and white blood cell counts were calculated with the Hmics package. String-DB 22 (version 12) was used to investigate biological terms and compartments associated with proteins of interest identified in the top-ranked proteins from the platelet proteome.

Results

Diversity of bleeding events across disease and treatment groups

Of the 201 hemato-oncological patients included in the control arm of the PREPAReS study, 176 met the inclusion criteria and had at least 1 sample available for proteomic analysis ( Figure 1 A). This cohort included patients diagnosed with AML (n = 74), MM (n = 33), NHL (n = 30), ALL (n = 16), MCL (n = 9), CLL (n = 1), and CML (n = 1). Additionally, 12 patients were classified with an undefined hemato-oncological disease and presented a variety of diagnoses including follicular lymphoma and mixed phenotype acute leukemia. Patients received different treatments, mainly autologous stem cell transplants (n = 76), remission induction chemotherapy (n = 60), consolidation chemotherapy (n = 22), allogeneic stem cell transplantation (n = 17), and maintenance (n = 1). The median number of platelet transfusions was 3 (interquartile range [IQR], 2-5), and these were most frequently administered to patients with AML (median, 5 [IQR, 3-8]) and least frequently to patients with MM (median, 2 [IQR, 1-3]). The average percentage of patients with a major bleed (WHO grade ≥2) was 19.3% across diagnoses, ranging from 37.5% in patients with ALL and 27% in patients with AML to no reported major bleeding events in patients with CML, CLL, and MCL ( Table 1 ).

To investigate the distribution and persistence of major bleeds, thrombocytopenia, and platelet transfusions in this cohort, we analyzed available daily records per patient, stratified by disease and treatment. Significant differences in the number of days with severe thrombocytopenia (platelet count of ≤10 × 10 9 /L; Figure 1 B) and the number of platelet transfusions ( Figure 1 C) received per patient were observed (Kruskal-Wallis, H(20) = 52.661 [ P < 9.113 × 10 −05 ] and 80.271 [ P < 3.532 × 10 −09 ], respectively; supplemental Table 1 A-B). No significant differences in the number of days with major bleeding ( Figure 1 D) were observed across groups with different diagnoses and treatments (Kruskal-Wallis P > .05), although we did observe higher number of days with major bleeding in patients with AML and ALL undergoing remission induction chemotherapy.

Experimental stability and proteomic depth in plasma samples

To identify proteins associated with bleeding severity in hemato-oncological patients, irrespective of diagnosis and treatment, we performed plasma proteomic analysis on 437 samples obtained from 176 eligible hemato-oncological patients, averaging 2 samples per patient. We quantified a median of 614 proteins (IQR, 595-627; supplemental Figure 2 A). To evaluate the analytical stability of this workflow, we included triplicates of a QC sample on each 96-well plate and determined CVs of repeated injections. CVs were consistent between 96-well plates ( supplemental Figure 2 B), with a median CV of 11% across all plates ( supplemental Figure 2 C). In total, 91% of proteins quantified in the QC samples had CVs <30% across the dynamic range ( supplemental Figure 2 D), showcasing analytical robustness across all measurements.

The association of plasma proteins and increased bleeding severity at the time of sampling

From the patient samples, 332 plasma samples corresponded to days without a bleed at the time of sampling, whereas 59, 43, and 3 samples corresponded to days with a WHO grade 1, grade 2, and grade 4 at the time of sampling, respectively. As a proxy of bleeding severity, these samples were divided into 3 ordinal categories based on the WHO grade scoring; no bleed, minor bleed (WHO grade 1), and major bleed (WHO grade ≥2). Principal component analysis revealed that global plasma profiles do not stratify patient samples based on their bleeding severity at the time of collection ( supplemental Figure 3 ). To further explore which individual proteins associate with an increased bleeding grade at the time of sampling, we implemented a protein-specific ordinal logistic regression model adjusted for demographic characteristics taking 1 sample per patient when the maximum bleeding grade was recorded ( Table 1 ). In this analysis, plasma levels of 55 proteins were significantly associated with bleeding severity (BH adjusted P < .05; supplemental Table 1 C; Figure 2 A). Abundance levels of 2 plasma proteins were positively associated (OR > 1) with bleeding severity, C-C motif chemokine 14 (CCL14), and dipeptidase 2 (DPEP2; OR, 2.3 [95% CI, 1.3-4.0] and 1.7 [95% CI, 1.2-2.5], respectively), although with large CIs ( Figure 2 B). Despite an observed trend of higher abundance levels of these proteins as severity of bleeding categories increased, this was not significant (Kruskal-Wallis P > .05; Figure 2 C). Abundance levels of 53 plasma proteins were negatively associated with bleeding severity (OR, <1; Figure 2 B). From these, the lowest point estimates were prenylcysteine oxidase 1 (PCYOX1), plasminogen (PLG), and coagulation factor VII (FVII) (OR, 0.14 [95% CI, 0.04-0.43], 0.19 [95% CI, 0.07-0.57], and 0.26 [95% CI, 0.12-0.54], respectively; Figure 2 D). For 45 of these 55 proteins, significant differences in the abundance levels between bleeding severity groups were observed ( supplemental Table 1 D-E; supplemental Figure 4 ). All these significant differences were between no bleeding and either minor or major bleeding groups, as observed in PCYOX1 and FVII (Kruskal-Wallis χ 2 test, H = 14.81 [degrees of freedom = 2; adjusted P = .0019]; and H = 11.141 [degrees of freedom = 2; adjusted P = .0059], respectively; Figure 2 E). Pairwise comparisons revealed that PCYOX1 protein levels from both bleeding groups (minor and major bleeding) were significantly different to the no bleeding group (Dunn test adjusted P = .003 for both comparisons) but not between them. Similar analysis for FVII protein levels showed significant differences when comparing major bleeding group and no bleeding group (Dunn test adjusted P = .003).

To understand the shared dynamics and highlight the potential underlying mechanisms, we grouped proteins based on expression patterns using correlation analysis. This revealed that 49 of 55 proteins with Pearson correlation coefficients ≥0.6 grouped in 4 clusters ( Figure 2 F). Three of these consisted of 2 or 3 proteins, which prevented us from performing enrichment analysis. These included PLG and shared regions from lipoprotein a (LPA) and PLG (cluster 1), shared regions of defensin α 1 and 3 (DEFA1 and DEFA3) and cysteine-rich secretory protein 3 (CRISP3; cluster 2), and FVII, vitamin K–dependent protein C (PROC), and retinol-binding protein 4 (RBP4; cluster 3). The fourth cluster contained 42 proteins enriched for biological processes including platelet aggregation and activation ( supplemental Table 1 F-G). Overall, this suggests that at the time of a bleeding event, hemostatic and platelet proteins are lower in abundance.

Plasma proteins at the first time point are associated with time to major bleed

Next, we investigated whether the levels of these 55 proteins at the first available plasma sample could be linked to the time to develop a major bleeding event. From the 176 available patients, 160 were eligible for time-to-event analysis, and 16 patients (9%) were excluded because they presented a major bleeding event (WHO grade ≥2) before their first available plasma sample ( Table 2 ; Figure 3 A). In total, 95 patients (54%) did not undergo a major bleed, whereas the remaining 65 (37%) suffered a major bleeding event, with a median time to event of 8 days (IQR, 4-13; Figure 3 B). A Cox proportional hazard model adjusted for demographic characteristics was performed with these 160 patients, taking into account the time to major bleed or the time to last follow-up censored at 30 days. Plasma levels of 9 of 55 proteins were associated with time to a major bleed, all with HR <1. These results suggest that increased abundance of these proteins in plasma are associated with longer times to a major bleeding event ( Figure 3 C; supplemental Table 1 H). Among these 9 proteins, protein disulfide-isomerase A3 (PDIA3) and PLG showed the lowest HR (HR, 0.63 [95% CI, 0.45-0.88] and 0.25 [95% CI, 0.09-0.64], respectively), followed by talin-1 (TLN1; HR, 0.69; 95% CI, 0.53-0.91), adenylyl cyclase–associated protein 1 (CAP1; HR, 0.75; 95% CI, 0.63-0.89]), LIM zinc finger domain–containing 1 (LIMS1; HR, 0.78; 95% CI, 0.66-0.92), Ras suppressor protein 1 (RSU1; HR, 0.81; 95% CI, 0.69-0.94), tropomyosin 2 (TPM2; HR, 0.81; 95% CI, 0.71-0.94), and shared regions (shared peptides) between proteins of TPM2, TPM3, and TPM4 (TPM2;TPM3;TPM4; HR, 0.82; 95% CI, 0.71-0.94). Next, we analyzed whether the median abundance of these proteins is able to stratify patients based on their time to a major bleeding event. Therefore, we separated patients into 2 categories: above and below the median protein abundances ( Figure 3 D; supplemental Figure 6 ). From these 9 proteins, PLG, RSU1, TPM2, and TPM4 showed no differences in the time to develop a major bleed when comparing these category groups. On the contrary, CAP1, LIMS1, PDIA3, TLN1, and TPM2;TPM3;TPM4 showed differences in the time to a major bleed between groups ( P <.05). This means that patients with protein levels below the median have a shorter time to a major bleed. All these 9 proteins are associated with platelets ( supplemental Table 1 I-J), indicating the potential relation of platelet protein levels in plasma with time to major bleeding events.

Platelet protein signature in plasma correlates with platelet counts but not platelet transfusions

To better understand the observed platelet protein signature, we investigated the correlation of these proteins with blood cell parameters, including platelet and white blood cell counts, in the first samples collected for each patient. Abundance levels of a total of 43 proteins correlated (ρ > 0.6; P < .001) with platelet counts ( Figure 4 A) but none to white blood cell counts ( Figure 4 B). Of these, 28 associated with bleeding severity, and 8 associated with both bleeding severity as well as time to major bleed ( Figure 4 C), with platelet factor 4 (PF4) and TPM2 showing the highest correlations, respectively ( Figure 4 D).

To account for the potential influence of platelet transfusions within this patient population, we looked into the number of platelet transfusions received before the first available sample for each patient. Of the 160 patients, 123 did not receive any transfusion, 31 received 1 transfusion, 4 received 2 transfusions, 1 received 3 transfusions, and 1 received 4 transfusions. We found no difference in time to a major bleed between patients who did or did not receive platelet transfusions ( Figure 4 E). Subsequently, we evaluated whether platelet transfusions had an effect on the levels of the top plasma proteins that correlated with platelet counts (PF4 and TPM2), as well as on the levels of proplatelet basic protein (PPBP) that we previously reported to increase after platelet transfusions. 23 No increase in plasma levels of PF4, TPM2, or PPBP was observed with a higher number of platelet transfusions ( Figure 4 F), suggesting that there is no correlation with the number of platelet transfusions received. Overall, abundance levels of the platelet proteins found to be associated with bleeding severity correlated with platelet counts but not with platelet transfusions.

Proteome signature linked to major bleeding risk has a cellular origin

To gain more insights into the characteristics of the platelet protein signatures and the associations with platelet count and bleeding severity, we performed an unbiased proteomic analysis of isolated platelets. For this, we built an in-house, in-depth reference platelet proteome with 4767 quantified proteins. In agreement with the enrichment analysis, 47 of 55 proteins that were associated with bleeding severity were identified in this platelet proteome ( Figure 5 A). Of these, 38 ranked among the top 500 most abundant proteins in platelets. These included 6 of the 9 proteins that were also associated with time to a major bleeding event ( Figure 5 B). Next, we investigated the functional pathways and cellular compartments enriched for these proteins. This analysis showed enrichment of proteins associated with alpha granules, actin filament bundle, and focal adhesion ( Figure 5 C). Interestingly, proteins associated with time to major bleed are related to general cellular structures, whereas proteins associated with bleeding severity are more related with platelet function. Taken together, this suggests that although bleeding severity is specifically related to platelets, time to major bleed may be more linked to a general aplasia.

Discussion

Patients with hematologic cancers frequently present with bleeding complications attributed to treatment-induced bone marrow aplasia. Studies have shown that neither platelet counts 8 nor other clinicopathological characteristics can identify patients at risk of major bleeds. 10 Here, we observed that lower abundance levels of coagulation and platelet proteins in plasma associated with bleeding severity in hemato-oncological patients and that abundance levels of a subset of these proteins is also linked to time to a major bleeding event.

We identified associations between bleeding events and the levels of 2 key vitamin K–dependent coagulation proteins in plasma: FVII and vitamin K–dependent protein C. We have previously shown a general dysregulation of liver-derived proteins, including coagulation FX, heparin cofactor 2 (SERPIND1), and FVII in patients with AML receiving supportive care. 12 Importantly, these results now suggest a protective association of FVII levels in plasma with bleeding severity (OR, <1). Interestingly, rare cases of post–stem cell transplantation–acquired FVII deficiency 24 , 25 , 26 and successful administration of recombinant FVII to treat major bleeding events after bone marrow transplantation have been reported, 27 , 28 although not consistently. 29 Overall, the observed low levels of FVII and other hemostatic parameters in cases of major bleeding in hemato-oncological patients need further exploration because they could be indicative of either protein consumption during active coagulation, alterations in overall liver protein synthesis, or protein deficiency after treatment.

In addition to coagulation proteins, we also observed a strong enrichment of platelet proteins associated with bleeding severity. This included proteins that are both platelet characteristic, such as PF4, ITGA2B, and GP1BB, as well as key regulators of cytoskeleton in proplatelet formation from megakaryocytes, such as TPM4 and myosin-9 (MYH9). 30 , 31 This platelet protein signature associated with bleeding severity prompts the query regarding the origin of these proteins.

Because all patients in this study received platelet transfusions, we first postulated that the origin of the platelet protein signature could be related to their supportive care. However, we did not observe a correlation with the number of transfusions before the first sample. This is in line with our previous report in which we only found an increase in PPBP upon administration of platelet transfusions in thrombocytopenic hemato-oncological patients. 23 These results therefore point toward platelet-derived extracellular vesicles 32 or immature platelets as the origins of the observed platelet protein signature. The percentage of reticulated platelets (RPs) that have recently been released from megakaryocytes in the bone marrow to the circulation has been postulated as a proxy for thrombopoietic activity in humans. Thus, RPs are suggested as a measurement of imminent platelet recovery in patients with chemotherapy-induced thrombocytopenia 33 and stem cell transplantation. 34 Hence, it could be possible that these RPs, characterized by increased activation, shape change, and cell adhesion, compared to non-RPs, 35 , 36 are the source of the platelet-protein signature. Furthermore, megakaryocytes and platelet aggregates have been reported in pulmonary arterial blood in human lung cancer cases, suggesting a role for extramedullar thrombopoiesis. 37 Therefore, potential sources for the signatures here described, including newly formed platelets and platelet-derived extracellular vesicles, should be explored in future studies.

Nine of the 55 proteins that associated with increasing bleeding grade were also related to the time to major bleeding events. Among these, we found proteins characteristic of cytoskeleton and cell adhesion. Some of them play a role in platelet activation and aggregation including TLN1 and PDIA3 (also known as Erp57), which are critical regulators of integrins 38 , 39 and CAP1, which regulates cytoskeleton in platelets upon activation. 40 This could further support the presence of functional platelets capable of activating and aggregating. However, many of these proteins are not exclusive of platelets, for example, RSU1 and LIMS1 are key interactors for cellular spreading and focal adhesion assembly, 41 , 42 which are ubiquitously expressed among blood cells. Therefore, the time to a major bleed could be associated with a general decrease in blood cells, suggesting that understanding the dynamics of recovery from different blood cell types may be important in the context of time to major bleeds in hemato-oncological patients.

There is a number of limitations to this study. Importantly, the included cohort was heterogenous in nature and small in size for underlying hematologic disease, which in combination with disease-specific treatment prevented adjustments for disease and treatment as covariates in statistical models. In addition, we were unable to adjust for known confounding factors such as renal dysfunction and coagulopathy due to missing data. Additionally, samples and bleeding records were matched on the day of sampling but not to the exact time for sample collection and bleeding event. As a result, we cannot distinguish between cause and consequence, and future studies should include standardized longitudinal samples to allow for a thorough investigation of the dynamics of these protein signatures surrounding bleeding.

In summary, our findings point toward a relevant role of coagulation factors and platelets in association with bleeding risk in hemato-oncological patients. The signatures observed here can imply consumption of coagulation factors and platelets due to their activation. Moreover, potential sources for the protein signatures identified here should be further explored, such as RPs and platelet-derived extracellular vesicles. In addition, the observation of a general cellular signature associated with time to major bleed, possibly due to general bone marrow aplasia, creates a need to further study the dynamics of cellular recovery and their influence on major bleeding events in these patients. Although the protein signatures observed here should be validated in future studies, our results are in agreement with previous reports on decreased FVII activity levels in hematologic malignancies during chemotherapy-induced thrombocytopenia. 43 Therefore, these results highlight the opportunity to improve clinical monitoring beyond platelet counts. Additionally, these findings may contribute to understanding the pathophysiology behind bleeding complications and thus, in the long term, may aid in tailoring current treatment protocols to specific patient needs, such as supplementation of coagulation factors. This could prioritize patients who might be at risk of bleeding complications, avoiding unnecessary platelet transfusions.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Pieter F. van der Meer died on 8 May 2025.

Acknowledgments

The authors acknowledge all the patients that participated in the Pathogen Reduction Evaluation and Predictive Analytical Rating Score (PREPAReS) trial. They thank Davina Ardanay, Stephanie Groot, and Jos Lorinser for sample retrieval from the PREPAReS study.

Terumo Blood and Cell Technologies supported funding for the PREPAReS study.

Authorship

Contribution: D.M.S. and E.R.S. performed plasma proteomics analysis, data analysis, designed figures, and wrote the manuscript; F.R.-R. advised and performed statistical analysis, designed figures, and reviewed and edited the manuscript; P.F.v.d.M., P.F.Y., and J.-L.K. contributed to study conception and design, clinical data collection, and the provision of study material from the Pathogen Reduction Evaluation and Predictive Analytical Rating Score; J.V. supervised research; M.v.d.B. designed study and supervised research; E.R.S., D.M.S., and M.v.d.B. interpreted data; P.L., A.J.H., E.R.S., and C.v.d.Z. developed mass spectrometry method and performed platelet proteomic analysis; T.J.P.v.D. advised on statistical analysis; A.J.H., T.J.P.v.D., J.V., and M.v.d.B. critically reviewed the manuscript; and D.M.S., E.R.S., F.R.-R., J.V., C.v.d.Z., P.L., A.J.H., T.J.P.v.D., J.-L.K., P.F.Y., and M.v.d.B. critically read the manuscript and approved the final manuscript.

Footnotes

D.M.S. and E.R.S. contributed equally to this work.

The data sets used and/or analyzed during this study are available in the ProteomeXchange/PRIDE archive database (identifier PXD071376). Individual participant data will not be shared.

The full-text version of this article contains a data supplement.

Supplementary Material

Supplemental Methods, Figures, and References

Supplemental Table

References

1. Stanworth SJ, Estcourt LJ, Powter G, et al. A no-prophylaxis platelet-transfusion strategy for hematologic cancers. N Engl J Med. 2013;368(19):1771–1780. doi: 10.1056/NEJMoa1212772.

2. Wandt H, Schaefer-Eckart K, Wendelin K, et al. Therapeutic platelet transfusion versus routine prophylactic transfusion in patients with haematological malignancies: an open-label, multicentre, randomised study. Lancet. 2012;380(9850):1309–1316. doi: 10.1016/S0140-6736(12)60689-8.

3. Ho G, Jonas BA, Li Q, Brunson A, Wun T, Keegan THM. Early mortality and complications in hospitalized adult Californians with acute myeloid leukaemia. Br J Haematol. 2017;177(5):791–799. doi: 10.1111/bjh.14631.

4. Wandt H, Frank M, Ehninger G, et al. Safety and cost effectiveness of a 10 × 109/L trigger for prophylactic platelet transfusions compared with the traditional 20 × 109/L trigger: a prospective comparative trial in 105 patients with acute myeloid leukemia. Blood. 1998;91(10):3601–3606.

5. Slichter SJ, Kaufman RM, Assmann SF, et al. Dose of prophylactic platelet transfusions and prevention of hemorrhage. N Engl J Med. 2010;362(7):600–613. doi: 10.1056/NEJMoa0904084.

6. Heddle NM, Cook RJ, Tinmouth A, et al. A randomized controlled trial comparing standard- and low-dose strategies for transfusion of platelets (SToP) to patients with thrombocytopenia. Blood. 2009;113(7):1564–1573. doi: 10.1182/blood-2008-09-178236.

7. Crighton GL, Estcourt LJ, Wood EM, Trivella M, Doree C, Stanworth S. A therapeutic-only versus prophylactic platelet transfusion strategy for preventing bleeding in patients with haematological disorders after myelosuppressive chemotherapy or stem cell transplantation. Cochrane Database Syst Rev. 2015;2015(9) doi: 10.1002/14651858.CD010981.pub2.

8. Uhl L, Assmann SF, Hamza TH, Harrison RW, Gernsheimer T, Slichter SJ. Laboratory predictors of bleeding and the effect of platelet and RBC transfusions on bleeding outcomes in the PLADO trial. Blood. 2017;130(10):1247–1258. doi: 10.1182/blood-2017-01-757930.

9. Friedmann AM, Sengul H, Lehmann H, Schwartz C, Goodman S. Do basic laboratory tests or clinical observations predict bleeding in thrombocytopenic oncology patients? A reevaluation of prophylactic platelet transfusions. Transfus Med Rev. 2002;16(1):34–45. doi: 10.1053/tmrv.2002.29403.

10. Cornelissen LL, Caram-Deelder C, Fustolo-Gunnink SF, et al. Expected individual benefit of prophylactic platelet transfusions in hemato-oncology patients based on bleeding risks. Transfusion (Paris) 2021;61(9):2578–2587. doi: 10.1111/trf.16587.

11. Ypma PF, van Geloven N, Kerkhoffs JLH, et al. The association between haemorrhage and markers of endothelial insufficiency and inflammation in patients with hypoproliferative thrombocytopenia: a cohort study. Br J Haematol. 2020;189(1):171–181. doi: 10.1111/bjh.16291.

12. Smit ER, Muñoz Sandoval D, Kreft IC, et al. Plasma proteomes of acute myeloid leukemia patients treated with transfusions reveal signatures of inflammation and hemostatic dysregulation. Transl Med Commun. 2024;9(1):27. doi: 10.1186/s41231-024-00189-5.

13. Van Der Meer PF, Ypma PF, Van Geloven N, et al. Hemostatic efficacy of pathogen-inactivated vs untreated platelets: a randomized controlled trial. Blood. 2018;132(2):223–231. doi: 10.1182/blood-2018-02-831289.

14. Ypma PF, Van Der Meer PF, Heddle NM, et al. A study protocol for a randomised controlled trial evaluating clinical effects of platelet transfusion products: the Pathogen Reduction Evaluation and Predictive Analytical Rating Score (PREPAReS) trial. BMJ Open. 2016;6(1) doi: 10.1136/bmjopen-2015-010156.

15. Kassambara A. rstatix: Pipe-Friendly Framework for Basic Statistical Tests. 2023 R package version 0.7.3.999.

16. Miller AB, Hoogstraten B, Staquet M, Winkler A. Reporting results of cancer treatment. Cancer. 1981;47(1):207–214. doi: 10.1002/1097-0142(19810101)47:1<207::aid-cncr2820470134>3.0.co;2-6.

17. Venables WN, Ripley BD. 4th ed. Springer US; 2002. Modern Applied Statistics with S.

18. Harrell F Jr. Hmisc: Harrell Miscellaneous. 2023 R package version 5.2-5.

19. Shannon P, Markiel A, Ozier O, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–2504. doi: 10.1101/gr.1239303.

20. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7) doi: 10.1093/nar/gkv007.

21. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. 1st ed. Springer; 2000.

22. Szklarczyk D, Kirsch R, Koutrouli M, et al. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 2023;51(D1):D638–D646. doi: 10.1093/nar/gkac1000.

23. Smit ER, Viegen I, van der Meer PF, et al. Impact of platelet transfusions on plasma proteomes in controlled endotoxemia and in hemato-oncological patients. Transfusion (Paris) 2025;65(10):1825–1838. doi: 10.1111/trf.18397.

24. Toor AA, Slungaard A, Hedner U, Weisdorf DJ, Key NS. Acquired factor VII deficiency in hematopoietic stem cell transplant recipients. Bone Marrow Transpl. 2002;29(5):403–408. doi: 10.1038/sj.bmt.1703381.

25. Anoun S, Lamchahab M, Oukkache B, Qachouh M, Benchekroun S, Quessar A. Acquired factor VII deficiency associated with acute myeloid leukemia. Blood Coagul Fibrinolysis. 2015;26(3):331–333. doi: 10.1097/MBC.0000000000000024.

26. Wolf EB, Li D, Fernandez AJ, et al. Acquired factor VII deficiency associated with chronic myeloid leukemia blast crisis. J Investig Med High Impact Case Rep. 2023;11 doi: 10.1177/23247096231209543.

27. Hoffman R, Eliakim R, Zuckerman T, Rowe JM, Brenner B. Successful use of recombinant activated factor VII in controlling upper gastrointestinal bleeding in a patient with relapsed acute myeloid leukemia. J Thromb Haemost. 2003;1(3):606–608. doi: 10.1046/j.1538-7836.2003.t01-5-00177.x.

28. Gerotziafas GT, Zervas C, Gavrielidis G, et al. Effective hemostasis with rFVIIa treatment in two patients with severe thrombocytopenia and life-threatening hemorrhage. Am J Hematol. 2002;69(3):219–222. doi: 10.1002/ajh.10056.

29. Blatt J, Gold SH, Wiley JM, Monahan PE, Cooper HC, Harvey D. Off-label use of recombinant factor VIIa in patients following bone marrow transplantation. Bone Marrow Transpl. 2001;28(4):405–407. doi: 10.1038/sj.bmt.1703157.

30. Pleines I, Woods J, Chappaz S, et al. Mutations in tropomyosin 4 underlie a rare form of human macrothrombocytopenia. J Clin Invest. 2017;127(3):814–829. doi: 10.1172/JCI86154.

31. Eckly A, Strassel C, Freund M, et al. Abnormal megakaryocyte morphology and proplatelet formation in mice with megakaryocyte-restricted MYH9 inactivation. Blood. 2009;113(14):3182–3189. doi: 10.1182/blood-2008-06-164061.

32. Vismara M, Manfredi M, Zarà M, et al. Proteomic and functional profiling of platelet-derived extracellular vesicles released under physiological or tumor-associated conditions. Cell Death Discov. 2022;8(1):467. doi: 10.1038/s41420-022-01263-3.

33. Wang C, Smith BR, Ault KA, Rinder HM. Reticulated platelets predict platelet count recovery following chemotherapy. Transfusion (Paris) 2002;42(3):368–374. doi: 10.1046/j.1537-2995.2002.00040.x.

34. Yang TH, Tsai CK, Wang HY, et al. Early prediction of platelet recovery with immature platelet fraction in patients receiving hematopoietic stem cell transplantation. Ann Hematol. 2024;103(11):4661–4670. doi: 10.1007/s00277-024-05951-1.

35. Hille L, Lenz M, Vlachos A, et al. Ultrastructural, transcriptional, and functional differences between human reticulated and non-reticulated platelets. J Thromb Haemost. 2020;18(8):2034–2046. doi: 10.1111/jth.14895.

36. Bongiovanni D, Santamaria G, Klug M, et al. Transcriptome analysis of reticulated platelets reveals a prothrombotic profile. Thromb Haemost. 2019;119(11):1795–1806. doi: 10.1055/s-0039-1695009.

37. Dejima H, Nakanishi H, Kuroda H, et al. Detection of abundant megakaryocytes in pulmonary artery blood in lung cancer patients using a microfluidic platform. Lung Cancer. 2018;125:128–135. doi: 10.1016/j.lungcan.2018.09.011.

38. Wang L, Wu Y, Zhou J, et al. Platelet-derived ERp57 mediates platelet incorporation into a growing thrombus by regulation of the αIIbβ3 integrin. Blood. 2013;122(22):3642–3650. doi: 10.1182/blood-2013-06-506691.

39. Petrich BG, Marchese P, Ruggeri ZM, et al. Talin is required for integrin-mediated platelet function in hemostasis and thrombosis. J Exp Med. 2007;204(13):3103–3111. doi: 10.1084/jem.20071800.

40. Joshi P, Riley DRJ, Khalil JS, Xiong H, Ji W, Rivero F. The membrane-associated fraction of cyclase associate protein 1 translocates to the cytosol upon platelet stimulation. Sci Rep. 2018;8(1):10804. doi: 10.1038/s41598-018-29151-w.

41. Ito S, Takahara Y, Hyodo T, et al. The roles of two distinct regions of PINCH-1 in the regulation of cell attachment and spreading. Mol Biol Cell. 2010;21(23):4120–4129. doi: 10.1091/mbc.E10-05-0459.

42. Fukuda K, Lu F, Qin J. Molecular basis for Ras suppressor-1 binding to PINCH-1 in focal adhesion assembly. J Biol Chem. 2021;296 doi: 10.1016/j.jbc.2021.100685.

43. Baaten CCFMJ, Moenen FCJI, Henskens YMC, et al. Impaired mitochondrial activity explains platelet dysfunction in thrombocytopenic cancer patients undergoing chemotherapy. Haematologica. 2018;103(9):1557–1567. doi: 10.3324/haematol.2017.185165.

Associated Data

Supplementary Materials

Supplemental Methods, Figures, and References

Supplemental Table