All adult kidney-transplant recipients followed in our hospital’s Department of Renal Transplantation were included, using an electronic medical record (R@N software) that automatically registers medical follow-up and blood-test results (French Data-Protection Authority [CNIL], decision 2009−413, number 1357154; July 2, 2009). We screened every patient with ≥1 blood sample(s) with platelets <100×10⁹/L between August 2002 (first year of R@N-software use) and November 2018. Patients whose PTTCP occurred <1-month posttransplant were excluded because PTTCP probably reflected surgical complications or anti-thymocyte globulin initiation, as were those with only 1 low platelet count or incomplete medical files. We then collected data from eligible patients’ files starting after the first month posttransplantation. For patients with ≥2 PTTCP episodes, only details of the first episode were considered. A second PTTCP episode was defined as occurring during follow-up after complete resolution of a first episode (after >3 months with platelets >100×10⁹/L).

The following information about the PTTCP episodes was collected: platelet-count nadir, another cytopenia also documented at any time during the PTTCP episode (defined as hemoglobin <10 g/dL or neutropenia <1.8×10⁹/L), splenomegaly (splenic cranial-caudal height >14 cm on imaging), any bone-marrow aspirate and/or biopsy histology findings (cell density, number of megakaryocytes, dysplasia of any lineage or blasts and bone-marrow biopsy morphology, when available) and, if possible, the peripheral/central mechanism able to explain the thrombocytopenia. Anti-platelet antibody and ¹¹¹indium-labeled platelet-scintigraphy results, only available for some patients, were also collected.

Anemia was defined as a hemoglobin level < 10 g/dL to highlight its significance given the various causes of anemia in renal transplant patients, particularly associated with impaired EPO secretion in chronic renal failure. The KDIGO guidelines recommend maintaining hemoglobin levels between 10 and 12 g/dL.

Patients were retrospectively assigned to 7 groups according to the underlying PTTCP etiology: 1) infection (concomitant pathogen presence), 2) drug induced (among drugs known to induce thrombocytopenia), 3) hypersplenism (splenomegaly >14 cm), 4) hematological disorder, 5) idiopathic thrombocytopenia purpura (ITP), 6) unknown cause, according to the main etiologies thought to induce thrombocytopenia, or 7) multiple causes, when ≥2 more were thought to independently trigger platelet decline simultaneously.

Among infectious causes, parvovirus B19 was incriminated only once, associating anemia with PTTCP, and only because the cytopenia resolved after intravenous immunoglobulin infusion (2). Sepsis-causing bacteria were retained when PTTCP resolved upon infection eradication. Cytomegalovirus (CMV) was deemed PTTCP-causative, when the infection was diagnosed according to international consensus guidelines (20), or PTTCP followed documented CMV infection with a normal bone-marrow smear during the first month of infection.

A therapeutic agent was held responsible when PTTCP resolved after withdrawal of a suspected drug. If several drugs were discontinued simultaneously and PTTCP resolved, all of them were considered PTTCP inducers.

Primary ITP was diagnosed according to international consensus guidelines (5). To simplify the analyses, CMV or other infectious agent-linked secondary ITP was considered to have only an “infection” etiology, not assigned to the ITP group.

We classified thrombotic microangiopathy-, hemophagocytic lymphohistiocytosis- or cirrhosis-inducing infections or drugs, as “hematologic disorder” or “hypersplenism” causes, not an infection or drug-induced PTTCP, including azathioprine-induced nodular regenerative hyperplasia (NRH).

For multiple-etiology or difficult-to-diagnose PTTCP episodes, a dedicated Adjudication Committee, comprised of nephrologists, immunologists and hematologists, determined the main cause(s), including all PTTCP episodes linked to ≥2 causes. Information on cardiovascular disease, graft rejection, renal failure and/or death were collected after PTTCP occurred; data collection was stopped after graft loss or for patients lost-to-follow-up.

Kidney transplant patients received preventive anticoagulation with heparin until day 4 post-transplant, after which the treatment was discontinued. During the thrombocytopenia episode, the ITP bleeding score ( Supplementary File S1 ) (19) was computed. It evaluates ITP-related bleeding severity, with a score >8 indicating hemorrhage and severe bleeding. Its items include the patient’s age, and various characteristics of cutaneous and mucosal (nose, throat, urinary, gynecologic, digestive or central nervous system) bleeding.

The need for packed red-cell transfusion(s) to counter bleeding or platelet transfusion(s), and any therapy given to raise the platelet count >100×10⁹/L were also recorded.

Categorical variables are expressed as absolute counts (percentages) and continuous variables as medians (interquartile [IQR] ranges) or mean (range). For descriptive analyses, categorical variables were compared with χ² or Fisher’s exact tests, as appropriate, and continuous variables with Student’s t-test. All statistical tests were 2-sided. Significance was defined as P<.05.

Graft and patient survivals post-PTTCP were estimated taking the beginning as PTTCP-episode onset, and the endpoint as the date of kidney failure, death or last known medical visit. The risk factors for severe bleeding were analyzed with Fine-Gray regressions, considering graft failure or patient death as competitive events, and a Cox proportional hazards model for those events.

Univariate analyses first selected risk factors for graft failure, bleeding and patient death as those achieving P<.2; they were included in multivariate models that were simplified by iterative backward elimination, retaining only the covariables with P≤.05.

From August 2002 to November 2018, 2118 patients were followed in our hospital after kidney transplantation. Among them, 240 in our database were identified as having platelet counts <100×10⁹/L; 51 were excluded, as defined above, and 189 (8.9%) patients were finally retained for this analysis ( Figure 1 ). Their characteristics at inclusion are detailed in Table 1 : median [IQR] age at first PTTCP was 57 [23-81] years, 125 (66%) were men, and median transplantation-to-PTTCP interval was 17.5 [1-624] months.

Among the 189 patients, 50 (26.5%) had isolated thrombocytopenia, 73 (38.6%) had PTTCP and anemia or neutropenia, and 66 (34.9%) had pancytopenia. Median platelet nadir was 51×10⁹/L [4-96]; nadirs were <10×10⁹/L for 9 (4.8%) patients, 11-50×10⁹/L for 78 (41.3%) and 51-100×10⁹/L for 102 (54%). Nadirs for 87 (46.0%) patients were <50×10⁹/L (i.e., 4.1% of all 2118 transplantees). Median PTTCP duration lasted 1 [2 days-204] months; the episode lasted <3 months for 129 (68.3%) patients and >12 months for 35 (18.5%). During follow-up, 151 (79.9%) PTTCP episodes resolved and 38 (20.1%) patients’ platelet counts never exceeded 100×10⁹/L or died before they could. Median follow-up from transplantation to death, graft failure or loss-to-follow-up was 7 [0-49] years. During that time, 25/189 (13.2%) patients had ≥2 PTTCP episodes.

Figure 2 summarizes the main identified causes of PTTCP. Notably, the Adjudication Committee identified ≥2 causes for 17 patients. The primary cause was drugs for 46, followed by infection for 38, unknown origin for 26, splenomegaly-related for 18 or hematologic disease for 25, and primary ITP for 18. One patient, not included in the figure, had only B9 and B12 deficiencies. Thirty-three (17.5%) patients experienced pretransplantation thrombocytopenia <100×10⁹/L and 13 (40%) of them relapsed with the same cause as previously.

CMV infection was incriminated in 34 (18%) patients: classed as the origin for 27 (79.4%) and 7 had multiple etiologies. Among the 34 kidney-transplant recipients with CMV infection, 26 had bone-marrow examinations during PTTCP; 21 (80.8%) of the latter had normal bone-marrow histology, suggesting increased platelet destruction. CMV-ITP often did not resolve after viral eradication; intravenous immunoglobulins or thrombopoietin-receptor agonists were successful for 7.

Azathioprine, prescribed to 37 PTTCP patients, was directly responsible for PTTCP in 16/37 (43.2%), notably for 8 with hepatic NRH. Other incriminated therapies were mycophenolate mofetil (MMF), valganciclovir, plasma exchange, intravenous immunoglobulins or linezolid (2 patients each) ( Table 2 ).

Among 13 thrombotic microangiopathies (including 2 with multiple etiologies), 10 were de novo: 4 were part of graft-rejection, 2 were calcineurin-inhibitor-induced, 1 had disseminated Bartonella henselae infection, 1 hemolytic uremic syndrome was linked to Escherichia coli and 2 others were atypical. Hemophagocytic lymphohistiocytosis was identified in 6 patients: 2 with posttransplantation lymphoproliferative disorders, 2 with sepsis, 1 with pulmonary tuberculosis, and 1 with disseminated Bartonella infection. Half of the 6 posttransplantation lymphoproliferative disorders were Epstein-Barr virus-related. In these patients, PTTCP was linked to bone marrow infiltration or chemotherapy, and one patient had splenomegaly.

PTTCPs appeared in 82 (43.4%) within year-1 posttransplantation. Among them, 25 (30.5%) were of infectious origin, including 22 (88%) CMV infections. After the first year, only 13/107 (12.1%) PTTCPs were infection-attributed and 6/107 (5.6%) CMV-related. Other causes did not differ between those 2 periods.

Specific interventions to raise the platelet count were started for 149/189 (78.8%) patients including immunosuppressive regimen dose-reduction for 97 (51.3%) patients, temporary/permanent withdrawal of a non-immunosuppressive drug for 51 (34.2%) and anti-infectious drug initiation for 49 (32.9%) ( Table 3 ).

Anti-platelet antibodies were detected in 16/57 (28.1%) tested patients, suggesting an autoimmune mechanism for the thrombocytopenia, even though this test lacks specificity and sensitivity (5). A bone-marrow smear was obtained for 121 (64.0%) patients, among whom 14 also had bone-marrow biopsies. Eighty-seven (71.9%) of the 121 patients had normal findings; notably, medullar cell density was normal or increased in 97 of them (80.2%). Because the thrombocytopenia mechanism was still unclear, platelet scintigraphy was done for 9 patients; it revealed shortened platelet life-span caused by increased platelet destruction in only 1 patient. Finally, although we agree making some diagnoses was not clear cut, those observations led us to conclude that hematopoiesis was affected in 54 (28.6%) patients, platelet destruction was enhanced in 128 (67.7%) and a mixed mechanism was involved in 7 (3.7%).

Sixty-two (32.8%) patients suffered bleeding complications when their platelet counts were <100×10⁹/L; bleeding was severe (defined as a bleeding score >8) in 40 (64.5%) of them. Six of (15%) those 40 patients died: 2 from hemorrhagic strokes, 3 from hemoptysis and 1 from digestive bleeding.

Patients experiencing ≥1 severe bleeding episode(s), compared to those without, respectively, had: lower platelet-count nadirs (40×10^9/L vs 53×10⁹/L; P=.005), higher percentages with ≥1 episode(s) of very low platelet counts <10×10⁹/L (5/40 [12.5%] vs 4/149 [2.7%]; P=.01), and requiring more packed red-cell transfusions (29/40 [72.5%] vs 4/149 [2.7%]; P=6×10^-24) and more platelet infusions (24/40 [60%] vs 22/149 [14.8%]; P=1.1×10^-8).

Patients who experienced at least one episode of thrombocytopenia, with a platelet count <50×10⁹/L, exhibited more severe bleeding compared to those with a platelet count >50×10⁹/L (29.9% versus 12.7%, p = 0.003). This trend was even more pronounced in patients with platelet counts < 10 × 10⁹/L, where the rates of severe bleeding were 12.5% compared to 2.7% (p = 0.01). Among those in the severe bleeding group, a greater proportion required cessation of anticoagulant or antiplatelet therapies compared to patients either not experiencing severe bleeding or with no bleeding incidents (20 out of 40 [50%] versus 18 out of 149 [12%], p < 0.00005). Univariate analyses ( Table 4 ) selected the following covariables as being significantly associated with a severe bleeding episode during PTTCP: pancytopenia, taking azathioprine at PTTCP onset and concomitant antiplatelet or anticoagulation therapy at that time. According to our multivariate analysis, severe bleeding during PTTCP was significantly and independently associated with pancytopenia and antiplatelet or anticoagulation therapy at that time.

During follow-up, graft rejection occurred in 34 (18%) patients, graft loss in 75 (39.7%) and 28 (14.8%) died. Mostly after the PTTCP episode resolved, 103 (54.5%) patients met the composite outcome of graft loss or death, within a median of 5.4 [0-49] years ( Table 3 ).

When comparing patients with subsequent graft loss or death to those without, the lowest recorded platelet counts were similar between the two groups (49.5×10⁹/L versus 52.1×10⁹/L; p = 0.55). Additionally, the proportion of patients with platelet counts <50×10⁹/L was comparable (48 out of 103 [46.6%] versus 37 out of 89 [41.6%]; p = 0.64). Univariate analyses ( Table 5 ) identified the following risk factors as being significantly associated with composite graft loss or death: donor and recipient ages, previous rejection episode before PTTCP, transplantation-to-PTTCP interval, PTTCP occurring at least 9 years posttransplantation, creatininemia at PTTCP diagnosis, bleeding during PTTCP and severe bleeding. According to our multivariate analyses, the risk factors significantly and independently associated with graft loss or death were: recipient’s age, transplantation-to-PTTCP interval, creatininemia at PTTCP onset and severe bleeding during PTTCP. Thirty-three patients (17.4%) experienced graft rejection following the PTTCP episode. Among these, 22 patients (64.7%) had a dose reduction of immunosuppressive drugs, primarily MMF, or an interruption of azathioprine, which occurred significantly more frequently (p<0.05) than in patients without graft rejection (26.4%).