Congenital Cyanotic Cardiac Surgery in Children: Is Algorithm-Based Point-of-Care Testing Essential to Prevent Bleeding?

Introduction

Patients with cyanotic congenital heart disease (CCHD) have multifactorial hematologic abnormalities such as thrombocytopenia, platelet dysfunction, and deficiency of coagulation factors.¹, ², ³, ⁴, ⁵, ⁶ These abnormalities are further accentuated due to hemodilution and platelet dysfunction during cardiopulmonary bypass (CPB).⁷, ⁸ This leads to increased bleeding tendency during surgery and postoperative period. Prediction of these hematologic abnormalities can identify patients at risk and optimize hemostatic therapy.

Conventional laboratory tests are not of much use during urgent situations because of prolonged turnaround time and inability to predict bleeding.⁹ Empirical treatment might lead to unnecessary blood product transfusion or suboptimal hemostatic management. Today, viscoelastic point-of-care (POC) methods such as thromboelastography (TEG), rotational thromboelastometry (ROTEM), and Sonoclot analyzer (Sienco Inc.) are available.¹⁰ The use of viscoelastic POC devices to guide hemostatic therapy has been demonstrated to decrease blood products requirement and improve clinical outcome.¹¹, ¹², ¹³, ¹⁴, ¹⁵, ¹⁶

We conducted this study to evaluate diagnostic accuracy of three different viscoelastic POC tests (TEG, ROTEM, and Sonoclot analyzer) to detect thrombocytopenia and hypofibrinogenemia in cyanotic cardiac surgery patients following an algorithm (AIIMS POC algorithm).

Material and Methods

Statistical Analysis

Continuous data are presented as mean and interquartile range (IQR). The POC parameters at baseline (T1) and 30 minutes after protamine administration (T2) were compared using Wilcoxon signed ranked test. Pearson's correlation coefficient was used to assess correlation between viscoelastic POC parameters and standard laboratory tests. Receiver operating characteristic (ROC) curve analysis was used to determine cutoff value, sensitivity, and specificity of POC parameters to detect thrombocytopenia (platelet count < 100,000/μL) and hypofibrinogenemia (serum fibrinogen concentration < 200 mg/dL). The optimal cutoff values of POC parameters were determined using Youden's index.¹⁸ A p < 0.05 was considered statistically significant. Statistical analysis was performed using computer software statistical package for social sciences (SPSS version 21.0, Inc.).

Results

The demographic characteristics and perioperative parameter of 105 patients included in this study are shown in Table 1. Patients undergoing CCHD repair including tetralogy of Fallot repair (n = 77), Glenn procedure (n = 9), Fontan procedure (n = 6), arterial switch over (n = 7), and total anomalous pulmonary venous connections (n = 6) were included. The mean (±SD) CPB time was 115.13 (±49.74) minutes. We observed statistically significant difference between POC parameters after anesthesia induction and 30 minutes after protamine administration indicating CPB-related hematologic changes (Table 2).

The correlations between baseline (T1) POC parameters and standard laboratory tests are shown in Table 3. We observed strong correlation between serum fibrinogen and ROTEM FIBTEM-A10 (r: 0.779; p <0.001) and FIBTEM MCF (r: 0.755; p < 0.001). The serum fibrinogen also showed moderate correlation with TEG α angle (r: 0.481; p < 0.001), TEG MA (r: 0.406; p < 0.001), and Sonoclot CR (r: 0.437; p < 0.001), respectively. Similarly, we found moderate correlation between Sonoclot PF (r: 0.513; p < 0.001) and platelet count. However, weak correlation was seen between platelet count and ROTEM EXTEM-A5 (r: 0.315; p < 0.001), EXTEM-A10 (r: 0.241; p < 0.05), and TEG MA (r: 0.239; p < 0.05). We observed poor correlation of platelet count with ROTEM INTEM MCF (r: 0.14; p = 0.157) and ROTEM-derived MCE (r: 0.11; p = 0.273).

Area under the curves (AUC) of ROC analysis was observed to evaluate the ability of ROTEM, TEG, and Sonoclot to detect thrombocytopenia and hypofibrinogenemia, whereas AUC (0.93) of ROTEM FIBTEM-A10 was found to be superior to other POC parameters for assessing hypofibrinogenemia (serum fibrinogen < 200 mg/dL) (Fig. 1). AUC (0.79) of TEG α angle, AUC (0.77) of TEG MA, and AUC (0.73) of Sonoclot CR were found to be comparable to each other.

Close

Fig. 1

Receiver operating characteristic curves of the ability of FIBTEM-A10, TEG α angle, TEG MA, and Sonoclot CR to predict hypofibrinogenemia (serum fibrinogen < 200 mg/dL). FIBTEM-A10 (AUC 0.93) is superior whereas TEG α angle (AUC 0.78), TEG MA (AUC 0.77) and Sonoclot CR (AUC 0.73) are comparable in predicting hypofibrinogenemia. A10, amplitude at 10 minutes; FIBTEM, fibrin-specific thromboelastometry; CR, clot rate; MA, maximum amplitude; TEG, thromboelastography.

Export to PPT

Whereas Sonoclot PF was found to have highest AUC (0.95) to predict thrombocytopenia (platelet count < 100,000/μL), AUC (0.70) of TEG MA was comparable to AUC of EXTEM A10 (AUC 0.70) (Fig. 2), the ROTEM-derived MCE parameter was found to have lowest AUC to predict thrombocytopenia (platelet count < 100,000/μL).

Close

Fig. 2

Receiver operating characteristic curves of the ability of EXTEM-A10, ROTEM MCE, TEG MA, and Sonoclot PF to predict thrombocytopenia (platelet count < 100,000/μL). Sonoclot PF (AUC 0.92) is superior to EXTEM-A10 (AUC 0.70), ROTEM MCE (AUC 0.64), and TEG MA (AUC 0.70) in predicting thrombocytopenia. A10, amplitude at 10 minutes; EXTEM, extrinsically activated thromboelastometry; MA, maximum amplitude; MCE, maximum clot elasticity; SONPF, Sonoclot platelet function.

Export to PPT

The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of baseline POC (ROTEM, TEG, Sonoclot) parameters to predict thrombocytopenia and hypofibrinogenemia are shown in Table 4. The cutoff values of baseline POC parameters to predict thrombocytopenia and hypofibrinogenemia were calculated from ROC curves using Youden's index.¹⁸ The ROTEM FIBTEM-A10 at threshold value ≤ 7.5 mm had highest sensitivity (87.2%) and specificity (80.3%) to predict hypofibrinogenemia. The Sonoclot PF at cutoff value ≤ 0.95 had highest sensitivity (100%) and specificity (83.7%) to predict thrombocytopenia. All POC parameters had high NPV and low PPV.

We formulated a POC algorithm based on cutoff value derived from ROC curves (Fig. 3). The most appropriate cutoff value was determined using Youden's index.¹⁷ We selected that value of the ROC curve as a cutoff value, where Youden's index (sensitivity + specificity-1) was maximal. The threshold value of EXTEM CT, INTEM CT, and TEG R is taken as 20% above the upper normal reference limit.¹⁹, ²⁰

Close

Fig. 3

Point-of-care (POC) algorithm using three viscoelastic POC devices; (ROTEM, TEG, and Sonoclot analyzer) to guide hemostatic therapy in cyanotic congenital heart disease patients. A10, amplitude at 10 minutes; ACT, activated clotting time; CR, clot rate; CT, clotting time; CDO, chest drain output; CRY0, cryoprecipitate; EACA, epsilon amino caproic acid; EXTEM, extrinsically activated thromboelastometry; FFP, fresh frozen plasma; FIBTEM, fibrin-specific thromboelastometry; ICU, intensive care unit; INTEM, intrinsically activated thromboelastometry; LY30, clot lysis after 30 minutes; MA, maximum amplitude; OT, operation theater; PC, platelet concentrate; PF, platelet function; R, reaction time; SON, Sonoclot; TEG, thromboelastography.

Export to PPT

Discussion

Viscoelastic POC instruments provide rapid assessment of coagulation cascade from beginning of clot formation to clot lysis. The use of POC-guided algorithm has been shown to decrease the requirement of blood product transfusion.¹⁴, ¹⁵, ¹⁶ These POC devices have shown good correlation with standard laboratory tests.²¹, ²², ²³

Ogawa et al conducted a prospective observational study to compare ROTEM-derived variables with standard laboratory tests in 26 patients, who were scheduled to undergo elective cardiac surgery. The author observed strong correlations between FIBTEM-A10 and fibrinogen level (r = 0.87; p < 0.001) and between EXTEM/INTEM-A10 variables and platelet count (r = 0.72 and 0.67, respectively; p < 0.001). ROC analysis demonstrated that EXTEM-A10 and INTEM-A10 are predictive of thrombocytopenia below 80 × 10⁹/L (AUC, 0.83 and 0.82, respectively), and FIBTEM-A10 was highly predictive of fibrinogen level below 200 mg/dL (AUC, 0.96).²¹

Saxena et al also assessed the correlations between Sonoclot variables and conventional coagulation tests. Clinical and laboratory data from 50 patients of four subgroups of liver disease, including decompensated cirrhosis, chronic hepatitis, cirrhosis with hepatocellular carcinoma (HCC), and acute-on-chronic liver failure, were analyzed. The authors observed significant positive correlation between Sonoclot CR and serum fibrinogen (r = 0.39; p < 0.004). The Sonoclot PF also showed strong correlation with platelet count (r = 0.62; p < 0.0001).²²

Espinosa et al conducted a prospective observational study in adult patients undergoing elective cardiac surgery with CPB. The author investigated possible correlations between TEG, ROTEM, Sonoclot variables, and routine hematologic coagulation tests. They observed significant correlations between FIBTEM MCF (r = 0.79; p < 0.0005), TEG MA (r = 0.76; p < 0.0005), and plasma fibrinogen levels. Similarly platelet count was demonstrated to correlate with TEG MA (r = 0.44, p = 0.009) and ROTEM EXTEM MCF (r = 0.53; p < 0.001). The authors did not study the correlation of Sonoclot PF with platelet count.²³

In our study also, we observed strong correlation between ROTEM FIBTEM-A10 and serum fibrinogen concentration (r = 0.78; p < 0.001) and moderate correlation between Sonoclot PF and platelet count (r = 0.51; p < 0.001). The TEG parameters such as TEG MA and TEG α angle showed statistically significant correlation with serum fibrinogen (r = 0.41; p < 0.001 and r = 0.48; p < 0.001). Platelet count also showed statistically significant correlation with EXTEM A5/A10 (r = 0.31; p = 0.001/r = 0. 24; p = 0.013) and TEG MA (r = 0.24; p = 0.015).

Olde et al investigated the ability of ROTEM to predict thrombocytopenia and hypofibrinogenemia in cardiac surgery patients using A5. The authors observed that FIBTEM-A5 can predict hypofibrinogenemia (fibrinogen < 200 mg/dL) with 100% sensitivity, 42% specificity, and ROC AUC 0.95. Similarly, EXTEM-A5 was found to predict thrombocytopenia (platelet count < 1,00,000/μL) with 100% sensitivity, 47% specificity, and ROC AUC 0.87.²⁴

In our study, we compared the ability of ROTEM, TEG, and Sonoclot analyzer to predict thrombocytopenia (platelet count < 100,000/μL) and hypofibrinogenemia (serum fibrinogen < 200 mg/dL) in patients undergoing repair of CCHD under CPB. We observed that while FIBTEM-A10 has strong ability to predict hypofibrinogenemia (ROC AUC, 0.93), Sonoclot PF has moderate ability to predict thrombocytopenia (ROC AUC, 0.92). In contrast, TEG α angle, TEG MA, and Sonoclot CR have moderate ability to predict hypofibrinogenemia (AUC; 0.79, 0.77, and 0.73, respectively). The ROTEM EXTEM-A10, MCE, and TEG MA were observed to have moderate ability to predict thrombocytopenia (AUC; 0.70, 0.64, and 0.69, respectively). The cutoff values of POC parameters to predict thrombocytopenia and hypofibrinogenemia were determined using ROC curves and Youden's index.¹⁸ We noted that FIBTEM-A10 at threshold value ≤ 7.5 mm has 87.2% sensitivity and 80.3% specificity to predict hypofibrinogenemia, whereas Sonoclot PF at threshold value ≤ 0.95 has 100% sensitivity and 83.7% specificity to predict thrombocytopenia. The difference in threshold value, sensitivity, specificity, and AUC noted in our study from previous study could be due to difference in studied population (cyanotic vs. acyanotic cardiac surgery patients). We propose a POC-guided algorithm using cutoff values of ROTEM, TEG, and Sonoclot parameters to guide hemostatic therapy in cyanotic cardiac surgery patients (Fig. 3). However, the efficacy of this algorithm should be studied in a randomized controlled trial.

Our study has few limitations. The platelet aggregometry POC tests were not performed in this study. This study was limited to cyanotic cardiac surgery patients; acyanotic cardiac patients were not included. Therefore, this study results might not be valid in acyanotic cardiac surgery patients. This study was planned to evaluate diagnostic accuracy of these viscoelastic POC tests to detect thrombocytopenia and hypofibrinogenemia. The efficacy of POC devices to guide hemostatic therapy and its effect on surgical outcome was not performed in this study. A randomized control study would be required to evaluate efficacy of these three POC tests on surgical outcome in CCHD repair.

Conclusion

In conclusion, although all three viscoelastic POC devices (TEG, ROTEM, and Sonoclot) can be used to detect hypofibrinogenemia and thrombocytopenia: ROTEM FIBTEM has highest diagnostic accuracy for hypofibrinogenemia, whereas Sonoclot PF has highest diagnostic value for thrombocytopenia in patients undergoing surgery for CCHD. In future, randomized controlled studies are required to evaluate efficacy of these tests on surgical outcome in CCHD patients. Use of these POC tests helps choose the right blood component therapy to reduce bleeding in these cyanotic children undergoing corrective surgery.