On a quest for fingerprints of crystalglobulinemia on platelets and endothelium

Key Points

• •Crystalglobulinemia is associated with a procoagulant platelet phenotype and high shear stress-like endothelial alterations.
• •Markers of platelet and endothelial function reflect monoclonal protein burden.

Visual Abstract

Abstract

Crystalglobulinemia is a rare plasma cell dyscrasia characterized by the crystallization of monoclonal proteins within the systemic vasculature and tissues, leading to occlusive vasculopathy, nephropathy, and tissue ischemia. We investigated platelet and endothelial activation status in a patient with crystalglobulinemia presenting with thrombotic events and acute kidney injury before and after anticlonal therapy. Platelet function assessment revealed the presence of procoagulant platelets, characterized by phosphatidylserine exposure and increased P-selectin expression in the absence of external stimuli. This phenotype was reversed by ex vivo calcium chelation. Furthermore, evidence supports FcγRIIA-mediated platelet activation by crystal complexes. Platelet proteomic analysis demonstrated upregulation of complement-related proteins, apolipoproteins, and proteins involved in platelet activation pathways. In addition, crystal proteomic analysis identified immunoglobulin heavy constant gamma-1 and albumin in a 1:2 ratio. Endothelial gene expression was evaluated using real-time polymerase chain reaction after coculturing the patient’s serum, isolated platelets, and citrated platelet-poor plasma with a serum-starved EA.hy926 endothelial monolayer. The patient’s serum increased endothelial gene expression of ICAM-1 and ADAMTS13, whereas VCAM-1 and E-selectin were significantly reduced compared with participants, a phenotype typically observed under high shear stress conditions. Platelet and endothelial dysfunction correlated with disease activity and demonstrated partial reversal after treatment, highlighting potential implications for disease monitoring and therapeutic targeting.

Introduction

Crystalglobulinemia, a rare complication of monoclonal gammopathies first described in 1972,¹ is characterized by deposition of crystallized monoclonal protein (M-protein) in the vasculature and tissues.^2,3 The mechanism of crystal formation in crystalglobulinemia is not fully understood. However, proposed mechanisms include abnormal N-glycosylation of light chains causing protein misfolding and insolubility, and a calcium-dependent immune complex formation of the M-protein with albumin.4, 5, 6 Intravascular crystal precipitation, often exacerbated by low temperatures, leads to occlusive vasculopathy primarily in microcirculation, without vasculitis.7, 8, 9 Histopathology often demonstrates intraluminal “pseudothrombi” formed by crystallized immunoglobulins and true fibrin deposition around crystals within thrombosed vessels, highlighting a distinct mechanism of vascular injury.8, 9, 10, 11, 12 Patients often present with crystalglobulin-associated nephropathy, retiform purpura, ulcers, neuropathy, and polyarthralgia.^10,11,13,14 Thrombotic complications in crystalglobulinemia include typically microvascular occlusion leading to renal failure and skin necrosis but may also involve larger vessels, causing venous and arterial thromboses. Reported cases describe intestinal ischemia, recurrent acute limb ischemia despite thrombectomy, deep vein thrombosis, and bilateral renal arterial thrombosis^7,8 unresponsive to thrombolysis. Plasma exchange reduces disease burden and may transiently improve symptoms; however, definitive treatment with anticlonal therapy is required even if SLiM-CRAB criteria for myeloma are unmet.¹² However, the mechanisms of thrombus formation in crystalglobulinemia-induced vasculopathy remain unexplored. This study evaluates platelet and endothelial activation, dominant players in thrombosis, in a patient with crystalglobulinemia-associated thrombotic events, and acute kidney injury.

Methods

This study involved a patient with immunoglobulin G (IgG)κ monoclonal gammopathy of clinical significance and kidney biopsy–confirmed crystalglobulinemia and 3 healthy participants.

Case presentation

A 47-year-old man, with a 3-year history of painful purpuric, nonblanching, nonpalpable rash, exacerbated at low temperatures, was diagnosed with having IgGκ monoclonal gammopathy of undetermined significance (MGUS). After 3 years from the initial symptoms, he presented with an unprovoked deep vein thrombosis involving the popliteal and tibial veins of the left leg and a segmental pulmonary embolism in the left lung. Superficial thromboses of the lesser saphenous vein and of superficial venous branches of the right foot were also observed. An ultrasound of the upper limbs also demonstrated bilateral thrombosis of the basilic veins. He was initially treated with therapeutic doses of low-molecular-weight heparin, followed by rivaroxaban 20 mg per day. After 1 year, he was urgently hospitalized for declining renal function with nonselective proteinuria. A brain computed tomography scan revealed pathological white matter lesions consistent with cerebral vasculitis. A kidney biopsy result established the diagnosis of crystalglobulinemia. The patient started treatment with daratumumab-bortezomib-cyclophosphamide-dexamethasone, followed by daratumumab maintenance (T1: at the time of diagnosis), achieving a partial response with clinical improvement (supplemental Table 1). Arterial duplex ultrasound of the right leg, performed for leg pain during effort, demonstrated occlusion of the tibioperoneal trunk with evidence of collateral vessel formation ∼12 months after start of therapy. After 20 months from first-line therapy initiation, the clinical syndrome relapsed with rising M-protein levels. Second-line therapy (T2: at the time of disease relapse) with pomalidomide-bortezomib-dexamethasone led to a minor response (<50% M-protein decrease) while improving clinical symptoms. He was further consolidated with high-dose melphalan and autologous stem cell transplant (ASCT). Although his M-protein level increased before ASCT, he achieved a very good partial response post-ASCT, with positive serum immunofixation and detectable bone marrow minimal residual disease detected by next-generation flow cytometry; his clinical syndrome is in remission.

Blood samples

Blood from the patient and 3 healthy participants was obtained after written informed consent by venous puncture from a cubital vein. Whole blood was collected in 2.7-mL 3.2% sodium citrate tubes (cWB) and 1.6-mL hirudin tubes (S-Monovette Hirudin; Sarstedt) (hWB) or 10-mL serum tubes with clotting activator. Time point definitions for blood sampling are detailed in supplemental Table 1.

Platelet function assays

Whole blood was collected in sodium citrate (cWB) and hirudin (hWB) tubes. Platelet aggregation was assessed via multiple electrode aggregometry (Multiplate; Roche Diagnostics, Mannheim, Germany).¹⁵ Heparin-induced multiple electrode aggregometry was performed using whole blood from healthy donors, which was incubated with the patient's platelet-poor plasma (PPP) and unfractionated sodium heparin, as previously described.¹⁶ Briefly, 340 μL of cWB was incubated for 1 minute with 200 μL of patients’ hirudinated PPP, citrated PPP (cPPP), or serum and 40 μL of saline or unfractionated heparin (UFH) (Leo Pharmaceutical Corp, Ballerup, Denmark) at final concentrations of 1 or 100 anti-Xa IU/mL. The experiment was repeated after preincubation of cWB at 37°C for 20 minutes with eculizumab (0.35 mg/mL; anti-human C5, humanized antibody, MedChemExpress), compstatin (0.12-0.18 mg/mL; C3 inhibitor, MedChemExpress), or the monoclonal antibody IV.3 (35 μg/mL; anti-FcγRII, STEMCELL Technologies, Canada), in the presence of UFH at 100 anti-Xa IU/mL. Platelet aggregation was recorded for 15 minutes.

Expression of the activated integrin αIIbβ3 (PAC-1), P-selectin (CD62P), CD63, and phosphatidylserine (PS) was determined before and after stimulation using a BD Accuri C6 Plus flow cytometer.¹⁷

Platelet isolation

Platelets were isolated from WB collected in acid-citrate-dextrose anticoagulant using a continuous density gradient (Sigma-Aldrich, St Louis, MO) according to the manufacturer's protocol.¹⁸ Residual leukocytes were further depleted using CD45-labeled DynaBeads (Invitrogen Dynal, Oslo, Norway) and a DynaMag-15 Magnet (Thermo Fisher Scientific) per the manufacturer's instructions.¹⁹

Crystal and platelet proteomic analysis

Crystals were dissected from patient’s bone marrow and WB smears with ZEISS PALM microbeam capture laser microdissection for mass spectrometry–based proteomics. Isolated platelets from patients and participants were lysed and digested using a modified SPEED protocol. Peptides were measured on a timsTOF fleX mass spectrometer (Bruker Daltonics) in library-free Data Independent Acquisition mode using Parallel Accumulation Serial Fragmentation. Raw files were preprocessed using DIA-NN version 1.8.2 (crystals) and FragPipe version 22.0 (platelets).²⁰

EA.hy-926 cells, real-time polymerase chain reaction for gene expression, and western blot analysis

Sera, isolated platelets, and cPPP from participants and patients were added to a serum-starved EA.hy-926 monolayer. After a 1-hour coculture, endothelial cell gene expression was evaluated using real-time polymerase chain reaction (supplemental Figure 1). Purified IgG was isolated from the sera of a healthy control donor, an MGUS patient, and the patient with crystalglobulinemia, using protein A/G magnetic beads (New England Biolabs). Then, 10 μg/mL of the purified IgG was treated for 24 hours with the EA.hy-926, and molecular analysis was performed via western blot. See the supplemental Material for further details on “Methods.”

All participants provided written informed consent in accordance with the Declaration of Helsinki. The study was conducted in accordance with ethical guidelines. Approval of our institutional review board was obtained.

Results

Assessment of platelet function

At diagnosis (T1_Pre) and at relapse (T2_Pre), platelet aggregation to all agonists was abolished in hWB but was normal in cWB as compared with control (T1_Pre is found in Figure 1A and T2_Pre is found in Figure 1B). Platelet aggregation in hWB was normal in patients with newly diagnosed cryoglobulinemia and IgGκ MGUS (Figure 1C).

Figure 1.

Impaired platelet aggregation in hirudin-anticoagulated whole blood from a patient with crystalglobulinemia. (A) Whole blood platelet aggregation tracings of a patient with newly diagnosed crystalglobulinemia (T1_Pre). Whole blood was anticoagulated with hirudin (hWB) and citrate (cWB). Tracings of platelet aggregation represent impedance changes plotted against time. Area under the curve (AUC) values, expressed in impedance units (U) are depicted in the white squares. Platelet response to arachidonic acid 0.5mM (ASPI test), ADP 6.5μM (ADP test), TRAP 32μM (TRAP test), and 0.77 mg/mL ristocetin (RISTO high test) were recorded for 6 minutes. In hWB, platelet aggregation to all agonists was abolished, whereas, in cWB, platelet aggregation was normal. Red and blue aggregation curves represent two parallel measurements which serve as internal control. (B) Bar graph of platelet aggregation in hWB and cWB in a patient with crystalglobulinemia and control participants. AUC values are expressed as mean values ± standard deviation. The patient’s samples were taken at three different time points before initiation of second-line therapy (T2_Pre), whereas the patient was in clinical and hematologic relapse. Three control participants were tested in parallel with the patient’s samples. Participants had lower AUC values in cWB compared with hWB, as expected. A marked decrease in patient’s AUC is observed in hWB compared with cWB. (C) Representative tracings of whole blood platelet aggregation performed in hirudin-anticoagulated whole blood in a patient with crystalglobulinemia before treatment initiation (i), a patient with cryoglobulinemia related to Waldenström macroglobulinemia evaluated at the time of initial diagnosis (ii), a patient with MGUS (iii), and a healthy control participant (iv). Tracings of platelet aggregation represent impedance changes plotted against time, expressed in impedance units (U). From top row to bottom: Platelet response to 0.5mM arachidonic acid (ASPI test), 6.5μM ADP (ADP test), 32μM TRAP (TRAP test), and 0.77 mg/mL ristocetin (RISTO high test) were recorded for 6 minutes. Suppression of whole blood platelet aggregation performed in hirudin-anticoagulated whole blood platelets was a unique finding in crystalglobulinemia and was not observed in participants without clinical or laboratory evidence for crystalglobulinemia. ADP, adenosine-diphosphate; ASPI, Aspirin test.

In hWB, unstimulated platelets demonstrated an abnormal, markedly high expression of CD62P (P-selectin) and CD63 along with spontaneous PS exposure (Figure 2A-B) as compared with the healthy participants. Moreover, a complete absence of PAC-1 expression, in thrombin receptor activating peptide (TRAP)–stimulated platelets was observed, corresponding to platelet aggregation failure to all agonists. In contrast, in cWB, we found a normal profile at baseline along with a lower expression of PAC-1, CD62P, and CD63 in TRAP-stimulated platelets compared with control.

Figure 2.

Spontaneous platelet activation in hirudin-anticoagulated whole blood evidenced by flow cytometry. (A) Flow cytometry histograms revealing mean fluorescence intensity (MFI) and percentage of CD41a⁺ platelets derived from citrated or hirudinated whole blood from the patient and control expressing activated platelet integrin αIIbβ3 (PAC-1), CD62P, CD63, and PS. Histograms depict the positive cells nonstimulated (baseline) or stimulated (with TRAP or calcium ionophore). (B) Bar graph with the percentage of expression of PAC-1, CD62P, CD63 before and after stimulation with TRAP, and PS before and after stimulation with calcium ionophore in hWB and cWB. Patient’s samples were taken before initiation of second-line treatment (T2_Pre) and were tested in parallel with samples from 3 healthy volunteers. In hWB, a marked spontaneous expression of PS, P-selectin, and CD63 and complete absence of TRAP-induced expression of PAC-1 were observed. (C) Flow cytometry histograms revealing MFI and percentage of cells in (I) nonstimulated and TRAP-stimulated CD41a⁺ platelets derived from 1 healthy donor’s and from the patient’s hWB at T2_Pre; (II) nonstimulated and TRAP-stimulated CD41a⁺ isolated platelets after incubation with Tyrode buffer or patient’s serum obtained at T2_Pre; (III) nonstimulated and calcium ionophore–stimulated CD41a⁺ platelets derived from 1 healthy donor’s and from the patient’s hWB at T2_Pre; and (IV) nonstimulated and calcium ionophore–stimulated CD41a+ isolated platelets after incubation with Tyrode buffer or patient’s serum obtained at T2_Pre. PLT, platelets.

To test whether the patient’s monoclonal protein (M-protein) was directed against the αIIbβ3 integrin, mixing experiments were performed using isolated platelets from a healthy control incubated with the patient’s serum. We found that the patient’s serum abolished PAC-1 binding on TRAP stimulation. It also induced PS exposure, reproducing the spontaneous exposure observed in the patient’s whole blood (Figure 2C).

Proteomic analysis of crystalglobulins

We then investigated whether the observed platelet phenotype could be associated with crystalglobulin formation. Crystals were visualized in blood smears prepared from serum, from the precipitate of nonanticoagulated whole blood after clotting and centrifugation, from both hWB and cWB before and after TRAP stimulation, and from the bone marrow (Figure 3A-G). Crystals observed in the precipitate, in nonstimulated hWB and cWB, and in bone marrow smears were rectangular in shape (Figure 3B-D,G). In contrast, in smears obtained from the reaction cuvette of TRAP-induced platelet aggregation in hWB and cWB (Figure 3E-F), crystals appeared as needle-shaped parallel arrays, partially disrupted and covered by amorphous material. Different crystal morphologies are reported in the literature, but the clinical significance of this finding remains unclear.²¹ Crystals from bone marrow smears and from nonstimulated hWB and cWB blood smears (Figure 3E-G) were dissected for proteomic analysis (Figure 3H). Their composition revealed immunoglobulin heavy-chain constant gamma-1 and albumin in ratios of 0.49 to 0.65 in hWB, 0.59 to 0.86 in cWB, and 0.49 in the bone marrow (supplemental Table 2). The immunoglobulin kappa constant to immunoglobulin heavy-chain constant gamma-1 ratio in crystals ranged from 0.08 to 0.16. Measurement of calcium concentration in the patient’s serum, hWB, and cWB was 10.8 mg/dL, 10.6 mg/dL, and 8.6 mg/dL, respectively, confirming the expected lower calcium concentration in cWB.

Figure 3.

Morphology and proteomic analysis of crystals in a patient with crystalglobulinemia. (A-G) Representative images of the crystal morphology. Crystals were visualized at the time of disease relapse, before T2_Pre, in smears prepared from various specimens. In the smears obtained from the reaction cuvette of TRAP-induced platelet activation, crystals appeared as needle-shaped parallel arrays partially disrupted and covered by amorphous material while the crystals in nonstimulated WB were rectangular in shape. (A) Unstained smear prepared from serum. (B) Giemsa-stained smear from the precipitate formed in nonanticoagulated whole blood on clotting and centrifugation. (C) Giemsa-stained smear of hWB. (D) Giemsa-stained smear of cWB. (E) Giemsa-stained smear from the reaction cuvette of TRAP-induced platelet aggregation in hWB. (F) Giemsa-stained smear from the reaction cuvette of TRAP-induced platelet aggregation in cWB. (G) Giemsa-stained smear of the bone marrow. (H) Experimental workflow for crystal characterization including dissection with laser capture microdissection, LC-MS–based proteomics. LMD, laser capture microdissection; LC-MS/MS, liquid chromatography-mass spectrometry/mass spectrometry.

Reevaluation during disease course

Platelet function was reevaluated at different time points throughout the disease course (Figure 4). Importantly, the platelet phenotype followed closely the patient’s clinical condition, M-protein levels, and hematologic response between different treatment lines. A decrease in M-protein, a hematologic response, and clinical improvement correlated with reversal of the abnormal platelet profile.

Figure 4.

Markers of platelet activation reflect disease activity in a patient with crystalglobulinemia. (A) Platelet reactivity assessed by hWB platelet aggregation correlates with disease activity. T1_Pre and T1_Post, before and after anticlonal treatment with daratumumab-bortezomib-cyclophosphamide-dexamethasone. T2_Pre and T2_Post, before and after T2 with pomalidomide-bortezomib-dexamethasone. TASCT_Pre and TASCT_Post, before and after ASCT. Red and blue line aggregation curves represent 2 parallel measurements which serve as internal control. At T1_Post, the patient was in partial response (>50% decrease in M-protein) and clinically improved and a transient increase in platelet aggregation was observed with increased responses to all agonists. As the patient started to deteriorate clinically, platelet reactivity decreased significantly, before T2 initiation. At T2_Post, response to all agonists increased transiently in correlation with a minor hematologic response. Before ASCT, the patient’s clinical condition deteriorated again, and monoclonal protein (M-protein) started to increase. A normal platelet response to all agonists was observed on day 100 post-ASCT when the patient was in very good partial response. (B) Bar graph platelet aggregation during disease course. AUC values expressed in U at different time points. Platelet response to 0.5mM arachidonic acid (ASPI test), 6.5μM ADP (ADP test), 32μM TRAP (TRAP test), and 0.77 mg/mL ristocetin (RISTO high test) were recorded for 6 minutes. (C) Flow cytometry histograms revealing MFI and percentage of cells expressing PAC-1, CD62P, CD63, and PS in nonstimulated or stimulated (with TRAP or calcium ionophore) CD41a⁺ platelets derived from hirudinated whole blood of the patient at T2_Pre, T2_Post, and TASCT_Post. The abnormal profile observed at diagnosis remained unchanged at T2_Post but was reversed at TASCT_Post except for a persistent increase of CD63 expression in nonstimulated platelets. (D) Bar graph with the percentage of expression of PAC-1, CD62P, CD63 before and after stimulation with TRAP, and PS before and after stimulation with calcium ionophore in hWB at before and after second-line treatment. Following ASCT, PAC1 expression increased after stimulation with TRAP, and spontaneous P-selectin and PS expression decreased dramatically. In contrast, CD63 remained abnormally increased in unstimulated platelets. ASCT, therefore, reversed platelet abnormalities observed before treatment except of a spontaneous CD63 expression. ADP, adenosine-diphosphate; ASPI, Aspirin test.

Mechanistic insights into the effect of crystals on platelet function

Proteomic analysis of the patient’s platelets at T2_Pre compared with platelets from healthy participants (n = 3) revealed a marked and significant increase in apolipoproteins (APOC3, APOA1, APOA4, APOB, APOL1, APOH) and complement-related proteins (C4A, C1QC, C4B, CFB,C1QB, CFH) (Figure 5A). Platelet Fcγ IgG receptor FcγRIIA was markedly lower at T2_Pre compared with healthy participants. Reassessment at disease T2_Post demonstrated higher FcγRIIA levels compared with T2_Pre and normalization of protein expression toward the healthy participant’s platelet proteomic profile (Figure 5B-D). Pathway enrichment analysis suggested activation of the complement and innate immune system in parallel with platelet activation and degranulation pathways (Figure 5E).

Figure 5.

Changes in platelet proteome and function in crystalglobulinemia before and after therapy. Volcano plots of the proteins with differential abundance (gene names) identified between the platelet proteome of (A) the healthy participants (n = 3) (control) and the patient at T2_Pre, (B) the control and the patient at T2_Post, and (C) the patient at T2_Pre and T2_Post. The black dots indicate upregulated or downregulated proteins, whereas the gray dots denote the protein abundance without marked differences. Gene names relative to the biological context are highlighted in red. The plot was created via FragPipe-Analyst.²² (D) Heat map illustrating the platelet proteomic profile of healthy participants (n = 3, control) and a patient before and after T2 with pomalidomide-bortezomib-dexamethasone (T2_Pre and T2_Post). The heat map represents differentially expressed proteins, with hierarchical clustering applied to highlight patterns of expression across conditions. Color intensity corresponds to relative protein abundance, with upregulated proteins shown in red and downregulated proteins in blue. The plot was created via FragPipe-Analyst.²² (E) Pathway enrichment analysis of the case platelets at T2_Pre vs control. Pathway enrichment was performed using the Reactome database within FragPipe-Analyst. Differentially expressed proteins identified between case platelets at T2_Pre and participants were analyzed for functional pathway enrichment. The top enriched pathways are shown, ranked by significance (˗log₁₀P value), with bar length representing pathway enrichment strength and color indicating false discovery rate. (F) Platelet aggregation tracings of Heparin-induced multiple electrode aggregometry (HIMEA). Patients’ cPPP and hirudinated PPP were mixed with whole blood from healthy participant in the presence of low and high concentrations of UFH or saline solution. In crystalglobulinemia, platelet aggregation in the presence of high UFH concentration was observed. (G) Effect of complement inhibition and FcγRII blockade on patient serum-induced platelet aggregation in HIMEA. Platelet aggregation induced by patients’ serum obtained during active disease (T2_Pre) in the presence of (UFH, 100 IU/mL) was reduced by treatment with the anti-C5 antibody eculizumab (0.35 mg/mL), the C3 inhibitor compstatin (0.12 and 0.18 mg/mL), or the FcγRII-blocking antibody IV.3 (35 μg/mL). Results are expressed as the percentage of residual platelet reactivity relative to aggregation observed with UFH 100 IU/mL alone.

We further sought to evaluate the role of FcγRIIA and complement in platelet activation. For this purpose, a functional assay for heparin-induced thrombocytopenia (HIT), a typical FcγRIIa-mediated thrombotic condition, was used. In the Heparin-induced multiple electrode aggregometry protocol, during active disease (T2_Pre), patients’ hirudinated PPP, cPPP in the presence of 100 IU/mL UFH induced platelet aggregation, a response that was not observed with clinically relevant UFH concentrations (1 IU/mL) or saline solution (Figure 5F). The addition of IV.3 (35 μg/mL) resulted in an 88% inhibition of the reaction, confirming FcγRIIa-mediated platelet activation. Platelet aggregation was also completely inhibited by the anti-C5 antibody eculizumab (0.35 mg/mL). Moreover, the C3 inhibitor compstatin reduced platelet aggregation by 30% at 0.12 mg/mL, with inhibition increasing to 78% at 0.18 mg/mL (Figure 5G).

Gene expression response of endothelial cells

Interaction of crystals with the endothelium may contribute to the thrombotic risk. Human endothelial cells’ gene expression was assessed in response to serum, cPPP, and platelets from the patient at 2 time points and the healthy participants (Figure 6). Exposure to the patient’s T2_Pre serum resulted in significantly higher expression of intracellular adhesion molecule-1 (ICAM-1) as compared with serum from healthy participants. Vascular cell adhesion molecule-1 VCAM-1 and e-selectin (SELE) expression were significantly blunted after treatment with patient’s T2_Pre serum (Figure 6A). Exposure to the patient’s platelets led to a marginal increase in ICAM1 and P-selectin (SELPLG) expression (Figure 6C). Although MGUS patient platelets also induced P-selectin expression, only platelets from our patient triggered concomitant ICAM1 upregulation (supplemental Figure 2).

Figure 6.

Patients' serum and platelets alter endothelial activation–related gene expression. Gene expression data for the evaluation of the effect of the patient and healthy volunteers’ on EA.hy-926 cells serum (A), cPPP, and (B) isolated platelets (100 platelets to 1 endothelial cell) (C). Gene expression was relative to the expression of the housekeeping gene actin, and data were analyzed using GraphPad Prism 8 software (GraphPad Software, Inc). One-way analysis of variance (ANOVA) was performed, and multiple comparisons among all means were conducted with Tukey post hoc test. P values <.05 were considered statistically significant and are reported accordingly. Data are plotted as means ± standard deviation. CCL2, C-C motif chemokine ligand 2 (MCP-1); CDH5, cadherin-5; F3, tissue factor; SERPINE1, plasminogen activator inhibitor 1; THBD, thrombomodulin; VWF, von Willebrand factor.

To determine whether the patient’s IgG is responsible for the observed gene expression in endothelial cells, we treated EA.hy926 endothelial cells with purified IgG preparations from a healthy volunteer, an MGUS patient, and the IgG of the index case for 24 hours. Treatment with purified IgG from the patient with crystalglobulinemia significantly increased NF-κB phosphorylation (Ser536) and E-selectin protein levels, without affecting ICAM-1 or VWF protein levels in comparison to control (Figure 7).

Figure 7.

EA.hy926 endothelial cell signaling on IgG treatment. Representative western blots and corresponding densitometric analysis of EA.hy926 endothelial cells treated for 24 hours with purified IgG preparations from a healthy volunteer, an MGUS patient, and the index patient with crystalglobulinemia. Treatment with the patient’s IgG significantly increased NFκB phosphorylation (Ser536) and E-selectin protein levels, without affecting ICAM-1 or VWF protein expression. Data are presented as mean ± SEM from n = 3 independent experiments. Statistical significance was determined by one-way ANOVA followed by Tukey post hoc test; P < .05 was considered significant and P values are depicted in the graph.

Discussion

This study provides the first insights into the interactions of crystallized immune complexes with platelets and the endothelium in a patient with MGUS-associated crystalglobulinemia. Ex vivo experiments assessing platelet and endothelial function were performed using specimens in which crystals were consistently visualized.

During active disease, findings from platelet aggregation and flow cytometry in hirudinated whole blood, but not in citrated whole blood, were consistent with the presence of procoagulant platelets. The platelet phenotype was characterized by PS exposure, increased P-selectin expression under unstimulated conditions, and blocked integrin αIIbβ3. Nevertheless, this phenotype was not accompanied by a hypercoagulable profile, as D-dimer levels remained normal, likely associated with the continuous anticoagulation administration of rivaroxaban to the patient throughout the observation period.

Our findings demonstrate a differential effect of the anticoagulant used in ex vivo experiments, suggesting that calcium concentration may regulate crystal formation and account for the distinct platelet activation profiles observed in citrated vs hirudinated whole blood. The role of calcium in stabilizing IgG binding sites and exposing specific antigenic determinants has been previously suggested.^6,23 Moreover, inhibition of crystallization by citrate has previously been demonstrated in vitro, although the exact mechanism remains unclear.²³ Notably, the effect of CaCl₂ on complex formation in vitro occurs over a relatively narrow concentration range, with enhanced precipitation at physiological levels and reduced complex formation at higher concentrations. The optimal calcium range for complex formation was not further investigated in our experimental settings, as this was beyond the scope of our study. Proteomic analysis of crystals visualized in whole blood and bone marrow was also consistent with prior reports indicating that the complex consists of IgG and albumin in an ∼1:2 ratio.^6,23

In total, although our methodology does not allow structural characterization of the crystals, the results support the hypothesis that calcium concentration may account for the discordance between hirudin-anticoagulated and citrated whole blood. Crystal formation in the serum is likely analogous to that in hirudin plasma, given the comparable calcium concentrations. Nonetheless, compositional differences between serum and plasma may differentially influence the outcomes of the ex vivo experiments.

In light of the above-mentioned results, we hypothesize that hirudin, by maintaining ex vivo calcium levels near physiological conditions, reveals the presence of procoagulant platelets through calcium-dependent crystal formation and provides a better reflection of the patients’ in vivo status. Citrate-induced inhibition of crystallization ex vivo has been reported^6,23 and may account for modulation of the ex vivo platelet phenotype through rapid dissociation of crystal complexes, thereby effectively switching off procoagulant platelet formation. It should be noted, however, that such citrate-mediated inhibition has not been demonstrated in vivo in our patient. Moreover, as evidenced in peripheral blood smears, crystals are also present in citrated whole blood, likely in smaller amounts compared with hirudinated WB, and not all assays are equally sensitive in detecting their presence.

The absence of PAC-1 binding on stimulation was also observed when the patient’s serum was mixed with control platelets. Moreover, the patient’s serum induced PS exposure in the absence of platelet agonists, indicating a shift toward the generation of procoagulant platelets. Previous studies have revealed that the M-protein can directly target αIIbβ3, resulting in severe platelet dysfunction. This may occur either through direct inhibition of ligand binding by monoclonal IgG²⁴ or via dysregulation of outside-in signaling by activating IgM.²⁵ Both mechanisms can manifest clinically as a bleeding syndrome, in contrast to the thrombotic phenotype observed in our patient. However, other markers of αIIbβ3 integrin expression and activation, including P-selectin expression, were not assessed in the mixing assay; thus, definitive conclusions cannot be drawn.

Interestingly, in this patient with crystalglobulinemia, evidence suggests FcγRIIA activation, which has been linked to procoagulant platelet generation.²⁶ This is supported by evidence from a functional assay for HIT, which revealed FcγRIIA-mediated platelet activation at high concentrations of UFH. Moreover, platelet proteomics revealed a decrease in FcγRIIA during active disease, which returned to levels comparable to those of a healthy participant after treatment. This was likely related to postactivation engagement of platelet FcγRIIA by crystal complexes, leading to internalization and degradation of FcγRIIA, as previously described.²⁷ Based on the HIT paradigm,²⁸ this finding suggests that crystalized M-protein may pathologically activate the FcγRIIA pathway when electrostatic coassembly of oppositely charged molecules occurs at an optimal ratio (ie, negatively charged UFH or albumin with positively charged calcium bound to M-protein). In contrast to HIT, the optimal ratio occurred at high UFH concentrations. Given that the formation of procoagulant platelets, as observed in our patient, has been reported as a critical mediator of thrombosis in other conditions associated with FcγRIIA-dependent platelet activation, such as HIT and vaccine-induced thrombocytopenia, it is plausible that the same mechanism applies in our patient.

Regarding complement involvement, in vitro evidence indicates that stimulated platelets can directly activate both the classical and alternative complement pathways, resulting in deposition of complement components on their surface.²⁹ Early classical pathway proteins, found in abundance during active disease in platelet proteomic analysis in our patient, may contribute to the opsonization of crystal complexes on platelets, analogous to HIT ultralarge immune complexes.³⁰ Moreover, blockade of complement components C3 and C5 markedly reduced platelet aggregation in the presence of high UFH concentrations, highlighting the contribution of complement activation to FcγRIIA-mediated platelet activation. However, further studies are needed to clarify the role of complement activation in the platelet phenotype observed in our patient.

Finally, the enrichment of apolipoproteins in the platelet proteome is of unclear origin. Most apolipoproteins are negatively charged, which raises the possibility that they may bind to the platelet surface through immune complex formation with IgG, similar to what has been described for albumin. Alternatively, their presence may reflect uptake from plasma, given the well-recognized ability of platelets to internalize circulating proteins through the open canalicular system, even after rigorous platelet isolation.³¹

Concerning the impact of crystalglobulinemia on the endothelium, our results suggest that during active disease, the patient’s serum shifts endothelial gene expression toward a high shear stress-like phenotype, an effect not observed with citrated plasma or in sera from patients with MGUS. Moreover, the patients’ endothelial phenotype reverted to normal after treatment. It has been proposed that shear stress increases ICAM-1 and decreases Vascular cell adhesion molecule-1 and E-selectin expression via tumor necrosis factor-α.³² The increase in ADAMTS13 expression during active disease could represent a compensatory mechanism to high shear stress-induced thrombogenicity, given that ADAMTS13-mediated cleavage of von Willebrand factor (VWF) is shear rate dependent.^33,34 However, as the multimeric composition of VWF was not assessed, the mechanism cannot be completely elucidated in this study. Differences between the patient’s cPPP and serum could be explained by the lower calcium concentration in cPPP, likely resulting in decreased crystal formation. However, differential effects related to clotting factors or inflammatory mediators between serum and plasma cannot be excluded. Although MGUS patient platelets also induced SELPLG expression, only platelets from our patient triggered concomitant ICAM1 upregulation, highlighting a key difference between MGUS and crystalglobulinemia that may be relevant to the patients’ thrombotic phenotype.

Treatment of endothelial cells with purified immunoglobulin increased NF-κB phosphorylation and E-selectin protein levels without affecting ICAM-1 or VWF, in contrast to the gene expression changes observed with the patient’s serum. These findings suggest that either the patient’s serum during the disease contains additional components that induce diverse endothelial cell responses or that differences in the quantity or composition of crystals between serum and purified IgG account for the observed effects. Nonetheless, both serum and purified IgG had an impact on endothelial cells, most likely through NF-κB activation and modulation of adhesion molecule expression.

Conclusions

Although our study included a single patient due to disease rarity, our findings support the calcium-dependent nature of crystalglobulins and reveal, for the first time, to our knowledge, their complex interaction with platelets and endothelium that tips the hemostatic balance toward a prothrombotic state. In a patient with MGUS-associated crystalglobulinemia, crystal complexes led to generation of procoagulant platelets enriched in complement components, key mediators of thrombus formation, and induced a high shear stress-like phenotype in the endothelium. Markers of platelet and endothelial function reflected M-protein burden, suggesting potential clinical implications for disease monitoring and therapeutic targeting. The longitudinal follow-up of the patient and the parallel course of M-protein levels with the clinical phenotype and the platelet profile support our observations. Further studies in additional patients are needed to better understand the complex pathophysiology of this rare entity.

Limitations

We acknowledge that, although IV.3 demonstrated FcγRIIA-mediated platelet activation, evidence linking FcγRIIA activation and formation of procoagulant platelets in our patient remains indirect, and further analysis of central signaling targets within this pathway was not performed. Moreover, the EA.hy926 endothelial cell line used in our experiments is immortalized and may not fully recapitulate the responses of primary human endothelial cells.

Conflict-of-interest disclosure: A.-M.D. reports honoraria from Takeda, Bristol Myers Squibb, Amgen, BeiGene, and Janssen. D.F. reports honoraria from Janssen. E.K. reports consultancy for Janssen, Pfizer, and Amgen; and honoraria from Janssen, Pfizer, Takeda, Genesis Pharma, and Amgen. The remaining authors declare no competing financial interests.

Ioanna Andreadou died on 13 January 2025.