Which cirrhotic patients are at high risk for bleeding during hospital procedures?

by Joseph Esherick, M.D., FAAFP, FHM

How do we assess a cirrhotic patients bleed risk prior to a planned bedside procedure?  Can our standard platelet count and coagulation studies, prothrombin time (PT/INR) and partial thromboplastin time (aPTT), accurately predict bleed risk in cirrhotic patients?  The quick answer to this is that these standard tests do NOT accurately predict bleed risk in cirrhotic patients for bedside procedures such as central line placement.

 

Liver dysfunction can affect levels of the procoagulant factors II, V, VII, IX, X and fibrinogen and the anticoagulant factors antithrombin, Protein C and Protein S.  In addition, patients with decompensated cirrhosis can have elevated levels of FVIII, vWF, and fibrinogen producing more of a hypercoagulable state.  To make things more complicated, 30-50% of patients with decompensated cirrhosis have systemic fibrinolysis because of the delayed clearance of fibrinolytic enzymes like tPA.  Significant fibrinolysis causes delayed bleeding and is suggested by a shortened euglobulin clot lysis time.  The difficulty is that cirrhotic patients with the same Childs-Pugh score can have variable levels of there procoagulant and anticoagulant proteins even though they have the same PT/INR and platelet levels.[1]

 

The prothrombin time/INR does not accurately predict bleed risk in cirrhosis like it does with patients on warfarin.  The international normalized ratio (INR) was developed to standardize PT reporting in patients on warfarin.  It has not been validated as a bleeding parameter in liver disease.  It is also variable from lab to lab depending on the different reagents utilized and therefore a patient with cirrhosis can have highly variable INR values with the same blood sample tested at different labs.[2]

 

Patients on warfarin with a therapeutic INR have levels of the Vitamin K-dependent clotting factors (II, VII, IX, and X) suppressed to below 30% normal.  On the other hand, patients with cirrhosis can have an INR>2 because FVII (half life 6 hours) is suppressed to levels less than 30%, but they can have normal levels of II, VIII, IX, and X.  In general, the bleed risk is low if patients have an elevated PT/INR and a normal aPTT.  The bleed risk is higher if the patient has an elevated PT/INR and an elevated aPTT more than 1.5 time the upper limit of normal; this suggests a deficiency in multiple clotting factors.  Unfortunately, the bleed risk suggested by the prolonged aPTT may be offset by elevated levels of FVIII and vWF meaning the procedural bleed risk for that patient would be low.  The bottom line is that our standard coagulation studies do not reliably predict the bleed risk in all cirrhotic patients.1

 

Other lab studies that can help to clarify the bleed risk in cirrhotic patients include checking a serum fibrinogen level, a platelet function analysis, and a thromboelastogram (if available).  Consider giving cryoprecipitate prior to a bedside procedure if the fibrinogen level is below 100 mg/dL.  A prolonged closure time with a platelet function analysis signifies a qualitative platelet dysfunction (if platelet counts exceed 50,000.  If the patient has a qualitative platelet dysfunction, administer desmopressin (DDAVP) 0.3 mcg/kg IV over 15 minutes and then perform the procedure after 30-60 minutes.[3]  A thromboelastogram (TEG) is a test of whole blood that assesses the time to clot and the clot lysis time.  If the patient has an increase reaction time, time until fibrin generation, FFP should be given.  If the patient has decreased fibrin build up and cross linking, cryoprecipitate should be given.  A decreased clot size suggests a qualitative platelet dysfunction when DDAVP may be useful.  Finally, a continually decreasing clot amplitude suggests fibrinolysis when transexamic acid (TXA) may be of benefit.

 

The studies on invasive procedures in patients with end-stage liver disease (ESLD) have examined the bleeding risk in percutaneous liver biopsy, intracranial pressure monitors, thoracentesis, and paracentesis.  Clinically significant bleeding occurs only 0.35-0.7% of the time with percutaneous liver biopsies.  These biopsies are felt to be safe if the prothrombin time is within 4 seconds of control, platelets >50,000, and bleeding time is less than 12 seconds.  A bleeding time greater than 12 seconds conferred a five-fold increased bleeding risk.[4]  The bleed risk for ICP monitor placement is 3-18% and there is no consensus on safe coagulation parameters for ICP monitor placement.  Major bleeding occurs in 0.2% of patients during central line placement and is generally considered safe if the aPTT is less than 1.5 times the upper limit of normal and platelets are >50,000.[5]  For paracentesis, major bleeding occurs only 0.2-1.2% of the time and is considered safe if the platelets are >50,000.[6]  The AASLD states that there is no level of INR elevation that would preclude the performance of a paracentesis.  Thoracentesis is generally safe to perform if the aPTT is less than 1.5 times the upper limit of normal and platelets are >50,000. [7]

 

In summary, our standard lab testing of platelets, PT/INR and PTT may not accurately predict bleed risk in cirrhotic patients.  Adding an analysis of the fibrinogen level, platelet function analysis and a TEG can better assess the procedural bleed risk in cirrhotic patients.  In general, bedside procedures (with possible exception of lumbar puncture) are safe to do with a platelet count >50,000 (assuming no qualitative platelet dysfunction), fibrinogen >100,000, and a normal aPTT level even if the PT/INR is marginally elevated.

Which cirrhotic patients are at high risk for bleeding during hospital procedures?

References:

[1] Crit Care Clin. 2005; 21: 563-587. Link.

[2] Clin Gastroenterol Hepatology. 2010; 8(10): 826-9. Link.

[3] Haemostasis. 1995; 25: 241-7. Link.

[4] NEJM. 2001; 344: 495-500. Link.

[5] Intensive Care Medicine. . 1999; 25: 481-485. Link.

[6] Hepatology. 2004; 40: 484-488. Link.

[7] Transfusion. 1991; 31: 164-171. Link.