CLINICAL AND LABORATORY OBSERVATIONS Protein C deficiency related to valproic acid therapy: A possible association with childhood stroke Ralph Gruppo, MD, Antonius DeGrauw, MD, Harold Fogelson, MD, Tracy Glauser, MD, Vinod Balasa, MD, and Peter Gartside, PhD We report a case of stroke in a child with acquired protein C deficiency receiving valproic acid (VPA). To investigate the possible association of VPA with protein C deficiency, protein C levels were measured in 20 children receiving VPA monotherapy and 20 children receiving other anticonvulsants. Protein C levels were reduced in up to 45% of the VPA-treated subjects. (J Pediatr 2000;137:714-8) Valproic acid has been demonstrated to cause alterations of hemostasis including abnormal bleeding time, qualitative and quantitative platelet abnormalities, decreased fibrinogen levels, and clinically significant bleeding.1-7 The effect of VPA on the natural anticoagulants protein C, protein S, and antithrombin has not been reported. An 18-monthold child had a stroke while receiving VPA therapy for a seizure disorder; she was found to have low plasma levels of protein C, which normalized after discontinuation of the anticonvulsant. From the Divisions of Hematology/Oncology and Neurology, Children’s Hospital Medical Center, Cincinnati, Ohio; and the Division of Biostatistics, University of Cincinnati Medical Center, Cincinnati, Ohio. Submitted for publication July 29, 1999; revision received Mar 30, 2000; accepted May 4, 2000. Reprint requests: Ralph A. Gruppo, MD, Division of Hematology/Oncology, Children’s Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229. Copyright © 2000 by Mosby, Inc. 0022-3476/2000/$12.00 + 0 9/22/108565 doi:10.1067/mpd.2000.108565 714 There have been no previous reports of thrombosis or stroke associated with VPA therapy. Therefore a study was undertaken to evaluate the levels of protein C and the other natural anticoagulant proteins and their possible association with VPA therapy in children. CASE REPORT The propositus is a 5-year-old girl who developed a seizure disorder at 2 months of age. Birth history was uneventful. An evaluation to determine the cause of the seizures was unrevealing. Head magnetic resonance imaging, performed at the ages of 4 months and 19 months, showed mild prominence of the ventricles, but findings were otherwise normal. Although she initially received phenytoin, she was switched to treatment with VPA after the development of an allergic rash. Development was delayed, but she was otherwise in good health. At 20 months of age, she presented with a seizure and right-sided hemiparesis clinically consistent with a stroke. A magnetic resonance imaging scan of the brain showed diffuse gyral swelling of the left hemisphere. The child was referred, at 24 months of age, for further evaluation. Physical findings at that time included persistent right-sided hemiparesis. Repeat magnetic resonance imaging of the brain revealed diffuse cortical volume loss on the left side, consistent with a previous diffuse ischemic injury. An evaluation to rule out an underlying prothrombotic disorder was performed at age 24 months. A complete blood count was normal except for a platelet count of 113 × 103/µL. A serum VPA level was 111 µg/mL. The fibrinogen level was 184 mg/dL. Plasma levels of antithrombin and protein S were normal. Results of tests for a lupus anticoagulant and anticardiolipin antibody were negative. Results of polymerase chain reaction testing for the Arg506 to Gln mutation of the factor V gene (factor V Leiden) and the 20210 G to A polymorphism in the 3´-untranslated region of the prothrombin gene were normal. The plasma protein C antigen level was 44% ELISA Enzyme-linked immunosorbent assay VPA Valproic acid (normal for age, ≥58%), and protein C anticoagulant activity was 30% (normal for age, ≥60%.) At age 28 months, protein C testing was repeated (antigen level, 50%; anticoagulant activity, GRUPPO ET AL THE JOURNAL OF PEDIATRICS VOLUME 137, NUMBER 5 28%). There was no family history of thrombosis, and protein C levels in both parents were normal. VPA was gradually discontinued by 35 months of age, and the child began receiving lamotrigine and carbamazepine. Studies at 39 months of age showed a rise in protein C levels to normal (antigen, 97%; anticoagulant activity, 73%). METHODS To determine the possible effects of VPA on the natural anticoagulant proteins, children who were followed up in the Division of Neurology at the Children’s Hospital Medical Center and who were receiving VPA as their only anticonvulsant medication were selected (group I) and compared with a similar group of children receiving other anticonvulsant therapy (group II). The study was approved by the Review Board on Investigations Involving Human Beings of the Children’s Hospital Medical Center. Appropriate informed consent was obtained for all participants. The age range of children eligible for the study was 6 months to 16 years. After a medical history and an inventory of current medications were taken, blood samples were obtained for the determination of a complete blood count and levels of VPA, alanine aminotransferase, serum albumin, fibrinogen, protein C, protein S, antithrombin, and D-dimer. Prothrombin time, activated partial thromboplastin time, and plasma levels of factor X antigen were measured to rule out possible effects of the anticonvulsants on the vitamin K–dependent coagulation proteins. Normal values of the anticoagulant proteins for age were derived from a previous study in 197 otherwise healthy children aged 6 months through 17 years (inclusive) who were admitted for elective sameday surgery. To define normal levels for age, children were divided into 4 non-overlapping groups: ages 6 months to 1 year, 1 to 5.5 years, 5.5 to A B C D Figure. Scatter graphs showing individual values for plasma concentrations of protein C antigen (A), protein C activity (B), fibrinogen (C), and whole blood platelet count (D) in children treated with VPA (group I) versus children treated with other anticonvulsants (group II). Circles represent normal values for age. Triangles represent abnormal values (≤5th percentile) for age. Normal values for protein C antigen and functional activity were derived from a study of healthy children for each of 4 non-overlapping age groups: 6 months to 1 year; 1 to 5.5 years; 5.5 to 11.2 years; and 11.2 to 16.9 years (see Methods). Shaded area represents normal adult range and is provided for reference only. 11.2 years, and 11.2 to 16.9 years. The cutoff for the lower limit of normal was set at ≤5th percentile for each age group. The study of healthy children was approved by the Review Board on Investigations Involving Human Beings, and appropriate informed consent was obtained. Normal values for adults were derived from a study of 40 healthy adults receiving no medication. Protein C antigen was determined by enzyme-linked immunosorbent assay with the Asserachrom Protein C kit (Diagnostica Stago, Asnieres-Sur, France). Protein C function was measured as chromogenic activity with the Coamatic Protein C kit (Chromogenix, Mölndal, Sweden). Protein S total and free levels were determined by ELISA with the Asserachrom kits (Diagnostica Stago). Antithrombin level was determined by the Spectrolyse antithrombin 715 GRUPPO ET AL THE JOURNAL OF PEDIATRICS NOVEMBER 2000 Table. Comparison of coagulation proteins in group I study subjects receiving VPA (n=20) compared with group II subjects receiving other anticonvulsants (n = 20) Protein C antigen Protein C function Protein S total Protein S free Antithrombin Group I (n = 20) Group II (n = 20) No. with abnormal values No. with abnormal values 9 (45%)* 8 (40%)† 0 0 2 (10%)‡ 0 0 0 1 (5%)‡ 0 The cut-off levels for the lower limit of normal for protein C, protein S, and antithrombin were derived from a study of healthy children and set at ≤5th percentile for each of 4 non-overlapping age groups: 6 months to 1 year; 1 to 5.5 years; 5.5 to 11.2 years; and 11.2 to 16.9 years (see Methods). *P = .001, Fisher exact test. †P = .002, Fisher exact test. ‡Not significant (P > .05, Fisher exact test). III chromogenic assay kit (Biopool, Burlington, Ontario, Canada). Factor X antigen level was measured by Laurell-type electroimmunoassay with polyclonal specific antibody (Celsus Laboratories, Inc, Cincinnati, Ohio). D-Dimer level was measured by ELISA with the Instant I.A. D-dimer kit (Diagnostica Stago). The distribution of the data was analyzed for normality. A Fisher exact test was used to compare the proportion of subjects in group I who had abnormal coagulation test results, which defined the contingency table, with subjects in group II. Correlations were analyzed by the non-parametric Kendall rank order test. RESULTS Group I The 20 subjects receiving VPA therapy for seizure disorders ranged in age from 6 months to 12 years (mean age, 6.8 years; median, 7.9 years). They had been receiving treatment for periods ranging from 2 months to 7 years (median, 18 months). No other anticonvulsants were being taken at the time the studies were performed. The only other medication being given was levo- 716 carnitine (2 subjects). None of the subjects were receiving antibiotic therapy. One subject had a history of meningitis, one had a history of prematurity and intraventricular hemorrhage, two had cerebral palsy, and one had Rett syndrome. None had known liver disease. One child was receiving feedings by nasogastric tube because of feeding problems and one had a history of recurrent upper respiratory tract infections. None of the study subjects had a history suggestive of venous thromboembolism or stroke. Group II The 20 subjects receiving anticonvulsant therapy other than VPA ranged in age from 2 to 12 years (mean age, 7.9 years; median, 7.9 years). One subject had a history of congenital cytomegalovirus infection, 2 had a history of lead poisoning, 4 had cerebral palsy, 2 had hydrocephalus with ventriculo-peritoneal shunts, and 1 had congenital heart disease, which had been surgically corrected. None of the subjects had received VPA therapy for at least 4 months. The numbers of subjects receiving various anticonvulsants are as follows: carbamazepine (n = 11), lamotrigine (n = 3), phenytoin (n = 2), topiramate (n = 1), lamotrigine and gabapentin (n = 1), lamotrigine and carbamazepine (n = 1), and lamotrigine and clonazepam (n = 1). Other medications included metoclopramide (n = 1) and imipramine (n = 1). None were receiving antibiotic therapy. VPA Levels, Platelet Counts, and Fibrinogen Levels Subjects in group I had VPA levels that ranged from 9.3 to 170 µg/mL (median level, 104 µg/mL). In all but one patient VPA dosage had been constant for at least 8 weeks before blood was obtained for this study. Fibrinogen levels were <150 mg/dL in 2 group I subjects (VPA levels 125 and 150 µg/mL). Platelet counts were decreased (41, 55, 44, 134 × 103/µL) in 4 group I subjects, 2 of whom had levels of VPA >100 µg/mL. One of the other 4 subjects with a low platelet count was admitted to the hospital for treatment of epistaxis and seizure, at which time the platelet count was 19 × 103/µL and VPA level was 256 µg/mL. VPA was withheld for 2 days. The platelet count increased to 41 × 103/µL, and the VPA level decreased to 25.6 µg/mL, at which time the subject was enrolled in the VPA study. No other subjects had a history of recent bleeding problems. All subjects in group II had normal platelet counts. One subject in group II had a fibrinogen level of 120 mg/dL. Protein C, Protein S, and Antithrombin Levels The levels of protein C antigen, protein C activity, fibrinogen, and platelets in groups I and II are compared in the Figure. An abnormal level for the coagulant and anticoagulant proteins was defined as ≤5th percentile for healthy children in the respective age group (see Methods). The numbers of subjects in groups I and II with abnormally low levels of the anticoagulant proteins are compared in the Table. Factor X levels were measured GRUPPO ET AL THE JOURNAL OF PEDIATRICS VOLUME 137, NUMBER 5 to rule out the effect of VPA on the vitamin K–dependent proteins. Normal levels were observed in all subjects in groups I and II. Normal alanine aminotransferase levels and normal prothrombin and activated partial thromboplastin times were observed in all subjects in both groups, providing additional evidence against vitamin K deficiency or liver dysfunction. DDimer levels were determined to rule out intravascular coagulation as a cause for decreased protein C levels. D-Dimer levels were undetectable in 18 of 20 VPA-treated subjects and borderline (0.5-1.0 µg/mL) in 2 subjects. In the group I subjects, protein C antigen levels tended to inversely correlate with VPA levels, but this association was weak (r = –0.27968, P = .0854). The protein C activity levels were more strongly correlated with VPA levels (r = –35545, P = .0295). DISCUSSION Patients receiving VPA have been reported to have decreased platelet counts, altered platelet aggregation, and prolonged bleeding times.1-5 The degrees of platelet count suppression and bleeding time prolongation correlate with VPA plasma concentrations.3,5 Hypofibrinogenemia has also been reported in association with VPA therapy, although the mechanism is unknown.6,7 Low fibrinogen levels have not been related to liver disease or disseminated intravascular coagulation. Increased fibrinolysis caused by VPA is unlikely, because elevated levels of fibrinogen degradation products or D-dimer have not been reported. A rapid rise in fibrinogen level follows discontinuation of the drug. We found reduced protein C levels in up to 45% of VPA-treated children. The association of VPA therapy and protein C deficiency has not been previously reported. Acquired protein C deficiency may be seen in liver disease, vitamin K deficiency states, thrombosis, or disseminated intravascular coagulation.8 The normal alanine aminotransferase levels, the normal coagulation study findings (prothrombin and activated partial thromboplastin time), and the normal levels of protein S and coagulation factor X provide evidence for normal liver function. Similarly, normal coagulation study results in association with normal levels of the vitamin K–dependent coagulation factors X and protein S provide evidence that vitamin K deficiency is not the cause of the low protein C levels. Normal levels of Ddimer were present in all but 2 patients receiving VPA therapy. The 2 patients with borderline elevations of D-dimer had normal protein C levels, suggesting that low-grade intravascular coagulation was not a cause of VPA-related protein C deficiency. Whether protein C deficiency secondary to VPA therapy is due to increased turnover or decreased production of the protein must await further studies. The protein C level in some of our study subjects was reduced to a level that has been reported to be associated with thromboembolic complications, including stroke, in children.9-11 Whether the reduced protein C level in our patient was causally related to her stroke, however, must remain speculative, because to date, there are no other reports of VPA therapy associated with thromboembolic disease. The lack of reports of thrombosis associated with VPA therapy is puzzling, considering the high incidence of protein C deficiency associated with its use. The apparent rarity of thrombosis associated with VPA therapy may be due to the failure to recognize the association. On the other hand, low levels of protein C may not be sufficient to induce thrombosis unless other prothrombotic factors are present. Particularly in children, thrombosis is a multicausal disease, and inherited and acquired risk factors are often simultaneously present.12 In some instances, the presence of multiple risk factors has supra-additive effects, with the relative thrombotic risk of the combined risk factors being greater than the sum of the relative risks of the individual factors.12 Besides deficiencies of antithrombin, protein C, and protein S, other common genetic risk factors for thrombosis include the factor V Leiden mutation and the 20210A mutation in the 3´-untranslated region of the prothrombin gene.12 The factor V Leiden mutation causes resistance to activated protein C. A thorough family history may aid in the recognition of the familial thrombophilic disorders. Common causes of acquired thrombophilic abnormalities in children include malignancy, infection, surgery, trauma, central venous catheters, and the presence of antiphospholipid antibody.10,11 We believe pediatricians should be aware of the possibility of hypercoagulability in children receiving VPA therapy, particularly in children with a family history of thrombosis. Unless other reports of VPA therapy associated with thromboembolic complications appear, we believe it is premature to suggest monitoring protein C levels during therapy. However, we would suggest an evaluation of protein C levels in any child suspected of having thrombotic complications while receiving VPA. Larger studies are needed to confirm our findings and further define the mechanism underlying acquired protein C deficiency related to VPA. We thank Ann Becker, BS, Ann Pillow, RN, and Davis Stroop, MS, for their technical assistance. REFERENCES 1. Smith J. Effects of valproic acid on platelets. Can J Hosp Pharm 1988;41: 93-4. 2. Ganick DJ, Sunder T, Finley JL. Severe hematologic toxicity of valproic acid. A report of four patients. Am J Pediatr Hematol Oncol 1990;12:80-5. 3. May RB, Sunder TR. Hematologic manifestations of long-term valproate therapy. Epilepsia 1993;34:1098-101. 4. Delgado MR, Riela AR, Mills J, Browne R, Roach ES. Thrombocytopenia secondary to high valproate levels in children with epilepsy. J Child Neurol 1994;9:311-4. 717 GRUPPO ET AL THE JOURNAL OF PEDIATRICS NOVEMBER 2000 5. Gidal B, Spencer N, Maly M, Pitterle M, Williams E, Collins M, et al. Valproatemediated disturbances of hemostasis: relationship to dose and plasma concentration. Neurology 1994;44:1418-22. 6. Dale BM, Purdie GH, Rischbieth RH. Fibrinogen depletion with sodium valproate [letter]. Lancet 1978;1:1316-7. 7. Ojemann LM, Wilensky AJ, Temkin O NR, Lovely MP. Fibrinogen and valproic acid [abstract]. Epilepsia 1981;22:242-3. 8. Miletich JP. Laboratory diagnosis of protein C deficiency. Semin Thromb Hemost 1990;16:169-76. 9. Israels SJ, Seshia SS. Childhood stroke associated with protein C or S deficiency. J Pediatr 1987;111:562-4. 10. Manco-Johnson MJ. Disorders of he- mostasis in childhood: risk factors for venous thromboembolism. Thromb Haemost 1997;78:710-4. 11. Nuss R, Hays T, Manco-Johnson M. Childhood thrombosis. Pediatrics 1995; 96:291-4. 12. Rosendaal FR. Venous thrombosis: a multicausal disease. Lancet 1999;353: 1167-73. N THE MOVE? Send us your new address at least six weeks ahead Don’t miss a single issue of the journal! To ensure prompt service when you change your address, please photocopy and complete the form below. Please send your change of address notification at least six weeks before your move to ensure continued service. We regret we cannot guarantee replacement of issues missed due to late notification. JOURNAL TITLE: Fill in the title of the journal here. OLD ADDRESS: NEW ADDRESS: Affix the address label from a recent issue of the journal here. Clearly print your new address here. Name Address City/State/ZIP COPY AND MAIL THIS FORM TO: Mosby Subscription Customer Service 6277 Sea Harbor Dr Orlando, FL 32887 718 OR FAX TO: 407-363-9661 OR PHONE: 800-654-2452 Outside the US, call 407-345-4000