Abstract
Objectives: The study aimed to identify factors influencing the severity of primary immune thrombocytopenia (ITP) during pregnancy, develop a predictive model for treatment response, and report maternal and neonatal outcomes associated with severe ITP. Design: A retrospective analysis was conducted on 155 pregnancies with severe ITP between January 2018 and April 2023 at a tertiary critical maternity referral center in Shanghai, China. Participants/Materials: The study included 155 pregnancies diagnosed with severe ITP, divided into groups based on the lowest platelet count (<30 × 109/L vs. 30–50 × 109/L) and first-line treatment response (non-response vs. response). Setting: The study was conducted at Renji Hospital, Shanghai Jiao Tong University School of Medicine, a tertiary critical maternity rescue referral center. Methods: Clinical characteristics and outcomes were compared between groups. A multivariable logistic regression model was used to identify factors associated with the severity of ITP. A prediction model for treatment response was established using LASSO-logistic regression and internally validated. Results: ITP severity was found to be correlated with low maximum amplitude of thromboelastography (OR 5.43, 95% CI: 1.48–16.00, p = 0.002), bleeding events (OR 4.91, 95% CI: 1.62–14.86, p = 0.005), and low reticulocytes (OR 2.40 × 10−7, 95% CI: 1.06 × 10−13 to 0.55, p = 0.04). Of the 118 patients who received first-line therapy, 52 (44%) responded. The dataset was randomly split into a training (N = 99) and test (N = 23) set with a ratio of 8:2. A predictive nomogram was created and internally validated showing good discrimination. The model yielded an area under receiver operating characteristic curve of 0.78 (0.69–0.87) and 0.85 (0.67–1.00) in the training and validation cohort, respectively. Earlier delivery and high rate of neonatal intensive care unit admission occurred with severe ITP and treatment failure. Limitations: The study was limited by a relatively small sample size and the retrospective observational design, which imposed limitations on the assessment of treatment efficacy. Conclusions: We identified clinical predictors of ITP severity and treatment resistance during pregnancy. A nomogram predicting first-line response was validated. These findings can facilitate clinical decision-making and counseling regarding this challenging pregnancy complication.
Introduction
Thrombocytopenia, defined as a platelet count below 150 × 109/L, complicates approximately 7–10% of pregnancies [1]. However, severe thrombocytopenia with platelet counts less than 50 × 109/L is quite rare, affecting less than 0.1% of pregnant women [2]. Consensus guidelines state that platelet counts exceeding 50 × 109/L are considered safe for both vaginal delivery and cesarean section [3]. Nonetheless, severe thrombocytopenia significantly increases the risks of maternal and fetal bleeding, necessitating close monitoring during pregnancy.
Primary immune thrombocytopenia (ITP) is an acquired autoimmune disorder characterized by accelerated platelet destruction and impaired production. ITP is the most common cause of severe thrombocytopenia in early pregnancy [4]. Evidence guiding the management of ITP during pregnancy is lacking as most data are extrapolated from studies in non-pregnant ITP patients. First-line treatment options include corticosteroids and intravenous immunoglobulin (IVIG) [5]. However, approximately 50% of ITP cases in pregnancy fail to respond to initial therapy [6]. Managing severe and refractory ITP in late pregnancy is particularly challenging as delivery approaches.
Given the rarity of severe thrombocytopenia in the pregnant population, few studies have identified clinical or laboratory factors predicting ITP severity or treatment outcomes. Such real-world data could significantly enhance management by facilitating risk stratification, treatment selection, patient counseling, and monitoring strategies. Therefore, our study aimed to (1) analyze clinical and laboratory variables associated with ITP severity during pregnancy, (2) develop and internally validate a predictive model for response to first-line treatment, and (3) report key maternal and neonatal outcomes related to this serious hematologic complication of pregnancy.
Materials and Methods
Study Population
A retrospective study was conducted with pregnancies diagnosed with primary ITP between January 2018 and April 2023 in Renji Hospital, Shanghai Jiao Tong University School of Medicine, a tertiary critical maternity referral center in Shanghai. This study was conducted in accordance with the principles of the Declaration of Helsinki and approved by the Renji Hospital Ethics Committee (LY2023-199-C). The requirement for informed consent was waived by the Institutional Review Board due to the retrospective nature of the study. All patient data were anonymized prior to analysis.
The diagnosis of ITP was made on the exclusion of other causes of thrombocytopenia according to the international consensus guidelines for primary ITP [7]. In this study, we included patients with severe thrombocytopenia defined as platelet count <50 × 109/L during any period of the pregnancy prior to delivery. Patients with thrombocytopenia secondary to other potential medical etiologies were excluded, which included preeclampsia, sepsis, hepatic disorders or autoimmune diseases. When thrombocytopenia was combined with anemia, bone marrow aspiration was further conducted to investigate hematological pathologies such as myelodysplastic syndrome, aplastic anemia, or hematologic malignancies. These criteria served to exclude cases of gestational thrombocytopenia, in which platelet counts remained above 70 × 109/L during pregnancy and normalized after childbirth.
We classified patients into two groups based on the lowest platelet count throughout the pregnancy: group A (below 30 × 109/L) and group B (30–50 × 109/L). All patients with severe ITP were evaluated and managed by a multidisciplinary team of hematologists and maternal-fetal medicine specialists, and the treatment decisions were through shared making between clinicians, patients and families.
Treatment was initiated when patients exhibited symptomatic bleeding, when platelet counts fell below 30 × 109/L, or to increase platelet counts to a level considered safe for procedures [5]. Regarding treatment strategies: (1) corticosteroids and IVIG served as the first-line treatment during pregnancy; (2) patients experiencing recurrent or persistent thrombocytopenia associated with bleeding after the first-line treatment were evaluated by hematologists and obstetricians for consideration of the second-line treatment, including immunosuppressants such as azathioprine and cyclosporin, and thrombopoietin receptor agonists; (3) platelet transfusions were reserved for temporary use to control life-threatening hemorrhage or to prepare a patient for urgent surgery or delivery [5]. Platelet transfusion was performed in cases with a platelet count below 50 × 109/L just before or during delivery [8].
A complete response was defined as an increase in platelet count greater than 100 × 109/L), while a partial response was defined as an increase in platelet count twice the baseline value but within the range of 30–100 × 109/L [6]. Based on the treatment effectiveness, pregnancies receiving the first-line treatment were further divided into two groups of non-response group and response group.
Clinical Characteristics
The medical records of all enrolled patients were independently reviewed and relevant data were retrieved: maternal demographics, medical and obstetric history, known diagnosis of ITP before pregnancy, gestational age at ITP diagnosis, bleeding events and treatment details, pre-delivery platelet transfusion, delivery outcomes (gestational age, mode of delivery, birth weight), postpartum hemorrhage, neonatal platelet counts at birth, neonatal intensive care unit (NICU) admission, and neonatal intracranial hemorrhage. Laboratory parameters were collected, which included serial blood counts (platelet, leukocyte, hemoglobin), reticulocyte count, thyroid function, serum ferritin, autoimmune antibodies, anti-platelet antibodies.
A history of thrombocytopenia was considered positive if there was the presence of a platelet count lower than 100 × 109/L before pregnancy. Bleeding events were defined via standard criteria from the International Society on Thrombosis and Hemostasis during pregnancy [9]. Postpartum hemorrhage was defined as ≥500 mL blood loss following vaginal delivery or ≥1,000 mL following cesarean section [10].
Statistical Analyses
Baseline characteristics were summarized using descriptive statistics. Continuous variables were expressed as mean ± standard deviation or median and interquartile range. Categorical variables were presented as number (percentage). Bivariate comparisons between groups utilized the χ2 test or Fisher’s exact test for categorical variables, and the Student’s t-test or Mann-Whitney U test for continuous variables, as appropriate. Multivariable logistic regression models were constructed to identify factors associated with ITP severity (i.e., group A and B). Covariates with p < 0.05 on univariable logistic regression were included in the final multivariable regression models. Then we developed and validated a model for predicting treatment effectiveness in patients with treatment data.
To fully use the sample, we employed multiple imputation by chained equation (MICE) to address missing data. We randomly split the cohort into a training set (N = 99) and a test set (N = 23) in a ratio of 8:2 to develop and validate a model for predicting treatment effectiveness (response vs. non-response). Least Absolute Shrinkage and Selection Operator (LASSO) was employed to select the candidate variables. The variables were fitted into a stepwise logistic regression model and a nomogram was established for clinical practice. Receiver operating characteristic (ROC) curve analysis was applied to evaluate the performance of the predictive model in the two datasets, respectively. Also, we assessed our model in certain key subpopulations (including age, gravidity, parturition, gestational age, body mass index, bleeding events, and ITP severity). Calibration curve analysis and decision curve analysis were also performed for the training and test sets, respectively. All analyses were performed using R statistics software (version 4.3.0). The MICE was performed using the “mice” R package. The threshold for statistical significance was defined as p < 0.05 (two-tailed).
Results
Clinical Characteristics
The patient selection flowcharts are shown in Figure 1. A total of 225 pregnancies with severe thrombocytopenia (platelet count <50 × 109/L) were included in this study. Of these, 33 pregnancies were excluded because of 11 cases of pregnancy-associated thrombocytopenia including gestational thrombocytopenia; hemolysis, elevated liver enzymes, and low platelet count syndrome; and fatty liver of pregnancy, 4 cases of secondary ITP, 5 cases associated with systemic conditions such as thrombotic thrombocytopenic purpura, hemolytic uremic syndrome, disseminated intravascular coagulation, and bone marrow disorders, 13 cases due to early pregnancy termination. Finally, 155 pregnancies with severe primary ITP were included, of whom 67.7% (105/150) were with a platelet count less than 30 × 109/L (group A), and 32.3% with a platelet count of 30–50 × 109/L (group B).
Of the 118 patients (76%) who received the first-line treatment, 42.3% received corticosteroid therapy (50/118), 7.6% received IVIG (9/118), 50% received corticosteroids combined with IVIG (59/118). Of those 118 patients following the first-line treatment, 44% showed a response (52/118) and 56% showed no response (66/118). Among the patients with no response to the first-line treatment, 26 initiated secondary treatment with the addition of thrombopoietin or eltrombopag, and only 2 cases showed good response to the secondary treatment. Out of the 155 included patients, 106 received transfusion of platelets (68.3%).
In group A (<30 × 109/L), 90% (94/105) received the first-line treatment, with 39 patients (41%, 39/94) showing an effective response. In group B (30–50 × 109/L), 48% (24/50) received the first-line treatment, with 13 patients (54%, 13/24) showing effective response.
Risk Factors Associated with ITP Severity
The characteristics of the population between group A (<30 × 109/L) and group B (30–50 × 109/L) were summarized in Table S1 (for all online suppl. material, see https://doi.org/10.1159/000541721). To explore potential risk factors associated with disease severity, univariate analysis was conducted, revealing a statistically significant correlation between ITP severity and reticulocyte count (OR 5.20 E−7, 95% CI: 4.76E−12 to 0.06, p = 0.01), the lowest counts of white blood cells (WBC, OR 0.76, 95% CI: 0.59–0.96, p = 0.004), the lowest hemoglobin (Hb, OR 0.97, 95% CI: 0.94–0.99, p < 0.001), bleeding events (OR 7.12, 95% CI: 1.94–26.06, p < 0.001), low maximum amplitude (MA) of thromboelastography (TEG) indicating low platelet-mediated clot strength (OR 8.52, 95% CI: 2.32–31.2, p < 0.001) (Table S2). In multivariate logistic regression analysis, low MA of TEG (OR 5.43, 95% CI: 1.48–16.00, p = 0.002), bleeding events (OR 4.91, 95% CI: 1.62–14.86, p = 0.005), and low reticulocytes (OR 2.40 × 10−7, 95% CI: 1.06 × 10−13 to 0.55, p = 0.04) were independent predictors for severity of ITP (Fig. 2).
Establishment of the Prediction Model for the First-Line Treatment Effectiveness
Following the assessment of first-line treatment effectiveness, patient characteristics between non-response and response were compared after MICE, as shown in Table 1. Autoimmune antibody, ferritin, lowest hemoglobin (Hb), and lowest WBC counts showed differences between non-response and response groups.
. | Non-response . | Response . | p value* . |
---|---|---|---|
N = 68 . | N = 54 . | ||
Age, years | 29.0 [28.0; 32.0] | 30.0 [26.0; 33.0] | 0.702 |
Gravidity | 0.906 | ||
0 | 32 (47.1%) | 27 (50.0%) | |
1 | 21 (30.9%) | 14 (25.9%) | |
2 | 8 (11.8%) | 8 (14.8%) | |
4 and 4+ | 7 (10.3%) | 5 (9.26%) | |
Parity | 0.685 | ||
0 | 50 (73.5%) | 37 (68.5%) | |
1 and 1+ | 18 (26.5%) | 17 (31.5%) | |
BMI, kg/m2 | 23.7 [21.5; 26.2] | 22.6 [20.6; 25.8] | 0.167 |
History of thrombocytopenia | 0.225 | ||
0 (N) | 16 (23.5%) | 19 (35.2%) | |
1 (Y) | 52 (76.5%) | 35 (64.8%) | |
Gestational age at diagnosis | 0.243 | ||
1 (1st trimester) | 22 (32.4%) | 25 (46.3%) | |
2 (2nd trimester) | 39 (57.4%) | 23 (42.6%) | |
3 (3rd trimester) | 7 (10.3%) | 6 (11.1%) | |
Severity of ITP | 0.272 | ||
0 (group B) | 11 (16.2%) | 14 (25.9%) | |
1 (group A) | 57 (83.8%) | 40 (74.1%) | |
Autoimmune antibody | 0.026 | ||
0 (negative) | 54 (79.4%) | 32 (59.3%) | |
1 (positive) | 14 (20.6%) | 22 (40.7%) | |
Reticulocyte count, % | 0.08 [0.06; 0.10] | 0.08 [0.07; 0.10] | 0.610 |
Bilirubin, μmol/L | 7.70 [4.95; 9.22] | 6.95 [5.20; 8.85] | 0.467 |
Coombs test | 0.583 | ||
0 (negative) | 67 (98.5%) | 52 (96.3%) | |
1 (positive) | 1 (1.47%) | 2 (3.70%) | |
Platelet antibody | 0.921 (not-tested excluded) | ||
0 (not tested) | 38 (55.9%) | 29 (53.7%) | |
1 (negative) | 16 (23.5%) | 13 (24.1%) | |
2 (positive) | 14 (20.6%) | 12 (22.2%) | |
TSH, mIU/L | 1.38 [0.89; 2.46] | 1.42 [0.76; 2.68] | 0.991 |
Ferritin, μg/L | 51.9 [29.4; 79.8] | 25.3 [17.1; 59.0] | 0.003 |
Low MA | 0.545 | ||
0 (negative) | 38 (55.9%) | 34 (63.0%) | |
1 (positive) | 30 (44.1%) | 20 (37.0%) | |
Bleeding events | 0.378 | ||
0 (N) | 35 (51.5%) | 33 (61.1%) | |
1 (Y) | 33 (48.5%) | 21 (38.9%) | |
Treatments | |||
IVIG | <0.001 | ||
0 (not used) | 25 (36.8%) | 24 (44.4%) | |
1 (response) | 0 (0.00%) | 16 (29.6%) | |
2 (non-response) | 43 (63.2%) | 14 (25.9%) | |
Corticosteroid | <0.001 | ||
0 (not used) | 5 (7.35%) | 4 (7.41%) | |
1 (response) | 0 (0.00%) | 48 (88.9%) | |
2 (non-response) | 63 (92.6%) | 2 (3.70%) | |
Secondary treatmenta | 0.002 | ||
0 (not used) | 46 (67.6%) | 50 (92.6%) | |
1 (used) | 22 (32.4%) | 4 (7.41%) | |
Lowest counts PLT, ×109/L | 18.0 [11.0; 25.0] | 21.0 [15.2; 29.5] | 0.064 |
Lowest Hb, g/L | 84.0 [69.0; 95.2] | 101 [91.2; 112] | <0.001 |
Lowest counts WBC, ×109/L | 3.94 [2.73; 5.21] | 6.28 [4.73; 7.37] | <0.001 |
. | Non-response . | Response . | p value* . |
---|---|---|---|
N = 68 . | N = 54 . | ||
Age, years | 29.0 [28.0; 32.0] | 30.0 [26.0; 33.0] | 0.702 |
Gravidity | 0.906 | ||
0 | 32 (47.1%) | 27 (50.0%) | |
1 | 21 (30.9%) | 14 (25.9%) | |
2 | 8 (11.8%) | 8 (14.8%) | |
4 and 4+ | 7 (10.3%) | 5 (9.26%) | |
Parity | 0.685 | ||
0 | 50 (73.5%) | 37 (68.5%) | |
1 and 1+ | 18 (26.5%) | 17 (31.5%) | |
BMI, kg/m2 | 23.7 [21.5; 26.2] | 22.6 [20.6; 25.8] | 0.167 |
History of thrombocytopenia | 0.225 | ||
0 (N) | 16 (23.5%) | 19 (35.2%) | |
1 (Y) | 52 (76.5%) | 35 (64.8%) | |
Gestational age at diagnosis | 0.243 | ||
1 (1st trimester) | 22 (32.4%) | 25 (46.3%) | |
2 (2nd trimester) | 39 (57.4%) | 23 (42.6%) | |
3 (3rd trimester) | 7 (10.3%) | 6 (11.1%) | |
Severity of ITP | 0.272 | ||
0 (group B) | 11 (16.2%) | 14 (25.9%) | |
1 (group A) | 57 (83.8%) | 40 (74.1%) | |
Autoimmune antibody | 0.026 | ||
0 (negative) | 54 (79.4%) | 32 (59.3%) | |
1 (positive) | 14 (20.6%) | 22 (40.7%) | |
Reticulocyte count, % | 0.08 [0.06; 0.10] | 0.08 [0.07; 0.10] | 0.610 |
Bilirubin, μmol/L | 7.70 [4.95; 9.22] | 6.95 [5.20; 8.85] | 0.467 |
Coombs test | 0.583 | ||
0 (negative) | 67 (98.5%) | 52 (96.3%) | |
1 (positive) | 1 (1.47%) | 2 (3.70%) | |
Platelet antibody | 0.921 (not-tested excluded) | ||
0 (not tested) | 38 (55.9%) | 29 (53.7%) | |
1 (negative) | 16 (23.5%) | 13 (24.1%) | |
2 (positive) | 14 (20.6%) | 12 (22.2%) | |
TSH, mIU/L | 1.38 [0.89; 2.46] | 1.42 [0.76; 2.68] | 0.991 |
Ferritin, μg/L | 51.9 [29.4; 79.8] | 25.3 [17.1; 59.0] | 0.003 |
Low MA | 0.545 | ||
0 (negative) | 38 (55.9%) | 34 (63.0%) | |
1 (positive) | 30 (44.1%) | 20 (37.0%) | |
Bleeding events | 0.378 | ||
0 (N) | 35 (51.5%) | 33 (61.1%) | |
1 (Y) | 33 (48.5%) | 21 (38.9%) | |
Treatments | |||
IVIG | <0.001 | ||
0 (not used) | 25 (36.8%) | 24 (44.4%) | |
1 (response) | 0 (0.00%) | 16 (29.6%) | |
2 (non-response) | 43 (63.2%) | 14 (25.9%) | |
Corticosteroid | <0.001 | ||
0 (not used) | 5 (7.35%) | 4 (7.41%) | |
1 (response) | 0 (0.00%) | 48 (88.9%) | |
2 (non-response) | 63 (92.6%) | 2 (3.70%) | |
Secondary treatmenta | 0.002 | ||
0 (not used) | 46 (67.6%) | 50 (92.6%) | |
1 (used) | 22 (32.4%) | 4 (7.41%) | |
Lowest counts PLT, ×109/L | 18.0 [11.0; 25.0] | 21.0 [15.2; 29.5] | 0.064 |
Lowest Hb, g/L | 84.0 [69.0; 95.2] | 101 [91.2; 112] | <0.001 |
Lowest counts WBC, ×109/L | 3.94 [2.73; 5.21] | 6.28 [4.73; 7.37] | <0.001 |
Data are presented as mean [95% Cl] or number (percentage).
BMI, body mass index; TSH, thyroid stimulating hormone; MA, maximum amplitude on thromboelastographic traces indicates platelet-mediated clot strength; IVIG, intravenous immunoglobulin; WBC, white blood cell; Hb, hemoglobin.
*p values are calculated by the Student’s t test or χ2 test.
aSecondary treatment including thrombopoietin or eltrombopag.
Patients were randomly divided into a training set (n = 99) and a validation set (n = 23). No statistically significant differences were observed between the two groups, except for reticulocyte count and lowest platelet counts (online suppl. Table S3). History of thrombocytopenia, negative autoimmune antibody, and lowest Hb were selected by LASSO (online suppl. Fig. S1). Of note, the lowest Hb was normally distributed; thus, it was fitted into the logistic model without log-transformation (online suppl. Fig. S2). Stepwise regression analysis demonstrated that the model with all three variables (i.e., history of thrombocytopenia, negative autoimmune antibody, and lowest Hb) was the most optimal. ROC curves were plotted in the training set and the validation set, respectively. The Area Under the Curve (AUC) of the training set and the validation set yielded 0.78 (95% CI: 0.69–0.87) and 0.84 (95% CI: 0.67–1.00), respectively (Fig. 3a). The prediction value (probability of response) was significantly higher among patients in the responder group in both the training (p < 0.001, Fig. 3b) and the validation set (p < 0.01, Fig. 3c), respectively.
We further constructed a nomogram based on our model to predict the probability of treatment effectiveness (Fig. 4). The calibration curves of the training set and validation set are shown in online supplementary eFigure 3. These illustrated the consistency between the nomogram prediction and actual observations. The calibration plot revealed good predictive accuracy of the nomogram. Decision curve analysis was then performed to analyze the clinical usability of the nomogram (online suppl. Fig. S4), indicating that this model demonstrated a favorable net benefit at a wide range of cutoff values. The subgroup analysis of ROC is shown in online supplementary Figure S5. The results of subgroup analysis show that the AUC of each subgroup ranges from 0.69 to 0.81, which demonstrates that this predictive model has good predictive performance in different subgroups.
As an example, for a pregnant woman with no history of thrombocytopenia, positive autoimmune antibody, and a lowest Hb of 79 g/L throughout the pregnancy, the total point on the nomogram would be 179, corresponding to a probability of treatment effectiveness of 60.3% (Fig. 4).
Maternal and Neonatal Outcomes
Maternal and neonatal outcomes are presented in Table 2. The overall incidence of postpartum hemorrhage in the study was 5.8% (9 of 155). The fetal loss occurred in 10 cases (6.4%), premature birth in 77 cases (49.7%), and cesarean section was performed in 131 cases (84.5%). Platelet transfusion was required in 106 cases (68.3%), and neonatal thrombocytopenia (defined as platelet count <150 × 109/L) was observed in 24 pregnancies (15.5%). Additionally, 49 newborns (31.6%) were transferred to NICU. There were no maternal deaths and no serious bleeding episodes observed.
. | Non-response (n = 68) . | Response (n = 54) . | p value . | Group A (PLT <30 × 109/L, n = 105) . | Group B (PLT30∼50 × 109/L, n = 50) . | p value . |
---|---|---|---|---|---|---|
Postpartum hemorrhage, n (%) | 1 (1.5%) | 7 (13.0%) | 0.021a | 7 (6.7%) | 2 (4%) | 0.393 |
Fetal loss, n (%) | 3 (4.4%) | 6 (11.1%) | 0.182 | 8 (7.6%) | 2 (4%) | 0.401 |
Premature birth, n (%) | 49 (72.1%) | 21 (38.9%) | <0.001a | 61 (58.1%) | 16 (32.0%) | <0.001b |
Caesarean section, n (%) | 63 (92.6%) | 38 (70.4%) | <0.001a | 87 (82.9%) | 44 (88.0%) | 0.410 |
PLT transfusion, n (%) | 65 (95.6%) | 21 (38.9%) | <0.001a | 75 (71.4%) | 31 (62.0%) | 0.244 |
Gestational age at delivery, week | 34.1 (3.1) | 36.6 (2.4) | <0.001a | 35.0 (3.0) | 36.7 (2.2) | 0.012b |
Birth weight, g | 2,441 (756) | 2,871 (642) | 0.030a | 2,605 (733) | 3,024 (626) | 0.011b |
Apgar (1 min) | 9.40 (1.36) | 9.71 (1.07) | 0.441 | 9.59 (1.18) | 9.73 (0.96) | 0.281 |
Apgar (5 min) | 9.90 (0.47) | 9.87 (0.66) | 0.140 | 9.92 (0.40) | 9.92 (0.51) | 0.392 |
Neonatal thrombocytopenia, n (%) | 8 (11.8%) | 12 (22.2%) | 0.061 | 14 (13.3%) | 10 (20.0%) | 0.403 |
Severe neonatal thrombocytopenia, n (%) | 3 (4.4%) | 1 (1.9%) | 0.430 | 3 (2.86%) | 1 (2%) | 0.753 |
NICU admission, n (%) | 34 (50.0%) | 13 (24.1%) | <0.001a | 39 (37.1%) | 10 (20.0%) | 0.019b |
. | Non-response (n = 68) . | Response (n = 54) . | p value . | Group A (PLT <30 × 109/L, n = 105) . | Group B (PLT30∼50 × 109/L, n = 50) . | p value . |
---|---|---|---|---|---|---|
Postpartum hemorrhage, n (%) | 1 (1.5%) | 7 (13.0%) | 0.021a | 7 (6.7%) | 2 (4%) | 0.393 |
Fetal loss, n (%) | 3 (4.4%) | 6 (11.1%) | 0.182 | 8 (7.6%) | 2 (4%) | 0.401 |
Premature birth, n (%) | 49 (72.1%) | 21 (38.9%) | <0.001a | 61 (58.1%) | 16 (32.0%) | <0.001b |
Caesarean section, n (%) | 63 (92.6%) | 38 (70.4%) | <0.001a | 87 (82.9%) | 44 (88.0%) | 0.410 |
PLT transfusion, n (%) | 65 (95.6%) | 21 (38.9%) | <0.001a | 75 (71.4%) | 31 (62.0%) | 0.244 |
Gestational age at delivery, week | 34.1 (3.1) | 36.6 (2.4) | <0.001a | 35.0 (3.0) | 36.7 (2.2) | 0.012b |
Birth weight, g | 2,441 (756) | 2,871 (642) | 0.030a | 2,605 (733) | 3,024 (626) | 0.011b |
Apgar (1 min) | 9.40 (1.36) | 9.71 (1.07) | 0.441 | 9.59 (1.18) | 9.73 (0.96) | 0.281 |
Apgar (5 min) | 9.90 (0.47) | 9.87 (0.66) | 0.140 | 9.92 (0.40) | 9.92 (0.51) | 0.392 |
Neonatal thrombocytopenia, n (%) | 8 (11.8%) | 12 (22.2%) | 0.061 | 14 (13.3%) | 10 (20.0%) | 0.403 |
Severe neonatal thrombocytopenia, n (%) | 3 (4.4%) | 1 (1.9%) | 0.430 | 3 (2.86%) | 1 (2%) | 0.753 |
NICU admission, n (%) | 34 (50.0%) | 13 (24.1%) | <0.001a | 39 (37.1%) | 10 (20.0%) | 0.019b |
p values are calculated by the Student t test, χ2 test.
Data are presented as mean (standard deviation) or number (percentage).
PLT, platelet; NICU, neonatal intensive care unit.
aSignificant difference between the non-response and response groups.
bSignificant difference between groups A and B.
Notably, the gestational week at delivery was significantly earlier in the non-response group compared to the response group, and corresponding significant differences were observed in premature birth and newborns’ weight. The rare of cesarean section rate in the non-response group was significantly higher than in the response group. Similarly, in the non-response group, a higher percentage of patients received platelet transfusions, and more newborns were transferred to NICU. Interestingly, response group showed higher rate of postpartum hemorrhage. No significant differences were observed between the groups regarding neonatal thrombocytopenia, fetal loss, and neonatal Apgar scores. When considering the impact of disease severity on pregnancy outcomes, it was found that group A (severe disease group) had higher rate of premature birth, lower birth weight, and a higher percentage of neonates transferred to the NICU compared to group B.
Discussion
While thrombocytopenia affects up to 10% of pregnancies, severe thrombocytopenia is a rare occurrence. In early gestations, most cases of thrombocytopenia requiring treatment are associated with ITP. Previous studies have primarily focused on maternal and fetal outcomes, rather than predictors of severity and treatment resistance [1, 2, 11‒16]. Our study helps fill this knowledge gap by identifying novel clinical and laboratory predictors of ITP severity and treatment resistance. We also developed and validated a predictive nomogram for assessing the response to the first-line treatment.
We found that impaired platelet aggregation (low MA) on TEG, bleeding events, and decreased reticulocyte counts were significantly associated with ITP severity. TEG offers a rapid functional assessment of overall coagulation status [17]. A low MA on TEG traces indicates compromised platelet-mediated clot strength. Our findings align with studies correlating TEG results with bleeding risk in thrombocytopenia [18], which underscores the clinical utility of TEG for evaluating coagulation disorders during pregnancy. Although TEG measurements were taken during the third trimester in our study, we recognize that TEG parameters can vary throughout pregnancy, with a trend toward hypercoagulability as gestation progresses. Thus, future studies should consider using gestational age-specific reference ranges for TEG parameters in pregnant women with ITP. Additionally, the presence of bleeding events was a predictor of more severe ITP, congruent with the known risks of hemorrhage with progressively lower platelet counts. Thus, clinical evidence of bleeding should prompt urgent evaluation. Although life-threatening bleeding was rare in our cohort, data suggest that intracranial and other severe hemorrhages may occasionally occur in the presence of profound thrombocytopenia [8]. Interestingly, the third predictor of severity was decreased reticulocyte counts, which reflect bone marrow hematopoietic function. This correlation aligns with the pathophysiology of ITP characterized by impaired aggregation and accelerated destruction of platelets. Hence, further investigation into the underlying mechanism is warranted.
During pregnancy, treatment is indicated for symptomatic bleeding, platelet counts below 30 × 109/L, or to raise platelets to a safe level before procedures. The first-line treatment options are corticosteroids and IVIG. A retrospective study of 98 singleton ITP pregnancies at two centers found effective responses in 18/47 (38%) treated with IVIG initially and 20/51 (39%) treated with steroids (p = 0.85) [6]. In our study, 52/118 (44.1%) responded to first-line therapy. We observed that prior thrombocytopenia, minimum gestational hemoglobin, and autoimmune antibody predicted treatment effectiveness. This novel finding highlights the value of comprehensive diagnostic testing. Evaluating autoimmune antibody and hemoglobin levels may provide insights into prognosis and guide management decisions. Our study validated a predictive nomogram incorporating these factors, providing a clinical tool for estimating an individual patient’s probability of responding to the first-line treatment.
In our study, 68.4% of pregnancies had a prior history of thrombocytopenia. Approximately half of ITP patients experienced significantly lower platelet counts during pregnancy, with 11% witnessing further dropping of platelet counts postpartum compared to pre-pregnancy levels [11]. Given the potential exacerbation of pregnancy with a history of ITP as well as the high likelihood of resistance to the first-line treatment, close monitoring and further diagnostic testing, such as platelet antibodies or even bone marrow aspiration, when appropriate, should be considered. Our research, in conjunction with that of others, indicates that a risk-stratified strategy for the surveillance and management of ITP during pregnancy is both suitable and effective. In cases where it is deemed necessary, a bone marrow biopsy should be conducted to differentiate between insufficient platelet production and accelerated platelet turnover in the assessment of pregnant women presenting with thrombocytopenia, although this is not a routine procedure [5]. Patients who exhibit severe or refractory ITP necessitate vigilant monitoring to ensure that their platelet levels are maintained within a safe range. Consequently, for pregnant patients with severe thrombocytopenia, it is advised that serum blood count monitoring be conducted every 2–4 weeks, or with increased frequency if they belong to the non-responder group or if their platelet count is declining [19].
Our study found that positive autoimmune antibodies, particularly antinuclear antibodies (ANAs), were a favorable predictor of response to the first-line treatment in pregnant women with ITP. This finding contrasts with some previous studies in non-pregnant populations. For instance, a prospective cohort study of 215 adults with primary ITP found no difference in response to the first-line treatment between ANA-positive and ANA-negative patients [8]. However, our results align with emerging evidence suggesting a complex relationship between autoantibodies and ITP. A plausible hypothesis for our observation is that ANA-positive ITP might represent a distinct subtype of the disease with a different underlying pathophysiology. The presence of ANAs could indicate a broader dysregulation of the immune system, potentially making these patients more responsive to immunomodulatory treatments like corticosteroids or IVIG. It is important to note that while ANAs are more prevalent in ITP patients compared to the general population [20], their role in the disease process remains unclear. Some studies have suggested that ANA positivity might precede the development of systemic autoimmune diseases [21]. For example, a retrospective study found that 7.7% of adult patients with primary ITP developed systemic lupus erythematosus over a mean follow-up of 6 years [22]. Our findings underscore the need for further research into the significance of autoantibodies in primary ITP, particularly in the context of pregnancy. Longitudinal studies examining the relationship between ANA status, treatment response, and long-term outcomes in pregnant women with ITP could provide valuable insights into disease mechanisms and guide more personalized treatment approaches.
Many clinicians have concerns s about peripartum hemorrhage in cases of severe ITP during late pregnancy. Nonetheless, reported postpartum hemorrhage rates in ITP vary widely, ranging from 8 to 13% [6, 8, 15]. In our cohort, the rate of peripartum hemorrhage (5.8%) was consistent with that reported in large population studies (3–6%). Regarding neonatal outcome, the severity of ITP and treatment response significantly impacted the premature delivery, birth weight and NICU admission. In our study, we found no difference in thrombocytopenia rates or median newborn platelet counts based on disease severity or gestational age at diagnosis. The most significant predictor was a previously affected sibling [16, 23]. In the context of neonatal thrombocytopenia in patients with severe ITP, our study indicates an incidence rate of 15.5%, which aligns with the broader spectrum of 9.5%–53% documented in the current medical literature [24‒27]. Severe neonatal thrombocytopenia, with platelets below 50 × 109/L was uncommon, and none of the affected newborns experienced severe bleeding complications or intracranial hemorrhage. The low risk of complications aligns with prior evidence of low rates of issues in newborns [16]. Our study not only elucidates factors associated with severe and refractory ITP during pregnancy but also challenges current practices regarding timing of delivery. The high rate (49.7%) of preterm birth, especially the iatrogenic premature birth rate of 67.5%, coupled with generally good maternal and neonatal outcomes, suggests that a more conservative approach to timing of delivery may be appropriate in many cases. This underscores the need for individualized management strategies based on comprehensive risk assessment rather than platelet count alone.
Limitations
A primary limitation of this study is the relatively small sample size, particularly in the test set for nomogram validation. Consequently, the model requires external validation in larger, multicenter cohorts to confirm its generalizability. The retrospective observational design also imposed limitations on the assessment of treatment efficacy. Patient selection bias and the absence of postpartum follow-up data were additional limitations. Nonetheless, our investigation provides real-world insights and identifies novel predictors to enhance prognosis and decision-making for this rare yet potentially life-threatening pregnancy complication.
Conclusions
We elucidated factors associated with severe and refractory ITP during pregnancy. In an era of precision medicine, predicting disease course and treatment response facilitates personalized management. Our findings should inspire prospective research to assess the generalizability of the predictive model and clarify the best practices for managing this complex clinical challenge.
Acknowledgment
We thank the multidisciplinary team for critical pregnant women in Renji hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China.
Statement of Ethics
Patient consent was not required as this study was based on publicly available data. This study was approved by the Ethics Committee of Renji Hospital Affiliated to Shanghai Jiao Tong University School of Medicine (LY2023-199-C). All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional and/or National Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
Conflict of Interest Statement
The authors declared that they have no conflicting interest.
Funding Sources
This work was supported by the Natural Science Foundation of Science and Technology Commission of Shanghai Municipality (No. 22ZR1438700).
Author Contributions
C.W., Y.Z., and N.Z. conceived and designed the study. C.W. and Z.H.H. collected data. C.W., Z.H.H., H.T.S., and N.Z. analyzed data. Z.H.H., H.T.S., and J.Y.W contributed to figures preparation. C.W., Y.Z., K.U.L., J.Y.W., and N.Z. interpreted the results. C.W., K.U.L., Y.Z., and N.Z. contributed to the writing of the manuscript. The author(s) read and approved the final manuscript.
Additional Information
Chuan Wang and Zhihong He contributed equally to this work.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. Further inquiries can be directed to the corresponding author.