- Email: [email protected]
Hepatic imaging in HELLP syndrome (hemolysis, elevated liver enzymes, and low platelet count)
Hepatic imaging in HELLP syndrome (hemolysis, elevated liver enzymes, and low platelet count) John R. Barton, MD," and Baha M. Sibai, MD b Lexington, Kentucky, and Memphis, Tennessee
OBJECTIVES: Our objective was to describe the hepatic imaging findings in selected patients with HELLP syndrome (hemolysis, elevated liver enzymes, and low platelet count) and to correlate these findings with the severity of concurrent clinical and laboratory abnormalities. STUDY DESIGN: Patients with laboratory criteria for HELLP syndrome with complaints of severe right upper quadrant abdominal pain in association with either shoulder pain, neck pain, or relapsing hypotension underwent imaging of the liver. Clinical and laboratory parameters were then correlated with the hepatic imaging findings. RESULTS: Thirty-four patients were evaluated in this study. Computed tomographic scanning of the liver was used for 33 patients. Additional imaging evaluations included magnetic resonance imaging for 4 patients and ultrasonographic evaluation of the liver for 5 patients. In 15 cases (45%) the computed tomographic results were abnormal. The most frequent abnormal hepatic imaging findings were subcapsular hematoma (n = 13) and intraparenchymal hemorrhage (n = 6). There was no statistically significant correlation between the presence of an abnormal hepatic imaging finding and the severity of liver function test abnormalities. However, the severity of thrombocytopenia did correlate with hepatic imaging findings (p = 0.04). In particular, an abnormal hepatic imaging finding was noted for 10 of 13 patients (77%) with a platelet count of <20 x 109/L (p = 0.012). CONCLUSIONS: Abnormalities in liver function test results do not accurately reflect the presence of abnormal hepatic imaging findings in HELLP syndrome. Patients with HELLP syndrome having complaints of right upper quadrant pain and neck pain, shoulder pain, or relapsing hypotension should undergo imaging of the liver. (Am J Obstet Gynecol 1996;174:1820-7.)
Key words: Preeclampsia, HELLP syndrome, computed tomography, magnetic resonance imaging
Hemolysis, abnormal liver function tests, and thrombocytopenia have been recognized as complications of preeclampsia-eclampsia for many years. In 1982 Weinstein 1 described 29 cases of severe preeclampsia-eclampsia in which the above complications occurred and coined the term HELLP syndrome, H f o r hemolysis, E L for elevated liver enzymes, and LP for low platelets. Patients with HELLP syndrome may have various signs and symptoms, although Sibai et al. 2 noted that 90% of patients with HELLP syndrome complain of epigastric or right upper quadrant pain. Right u p p e r quadrant or epigastric pain is thought to result from obstruction of blood flow in the hepatic sinusoids because of large hyaline deposits of fibrin-like material associated with periportal or focal pa-
From the Department of Obstetrics and Gynecology, Central Baptist Hospital, ~ and the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Tennessee, Memphis. b Annual Central Prize Award, presented at the Sixty-third Annual Meeting of The Central Association of Obstetricians and Gynecologists, Palm Desert, California, October 19-21, 1995. Reprint requests: John R. Barton, MD, Perinatal Diagnostic Cent~ Central Baptist Hospital, 1740 Nicholasville Road, Lexington, KY 40503. Copyright © 1996 by Mosby-Year Book, Inc. 0002-9378/96 $5.00+ 0 6/6/72246 1820
renchymal necrosis? -5 These histopathologic findings may be related to the elevated liver enzymes commonly observed in patients with this syndrome. In certain cases the cellular necrosis and hemorrhage may be severe enough to result in liver infarction or formation of a subcapsular hematoma. 6' 7 The purpose of this study was to describe the hepatic imaging findings in selected patients with HELLP syndrome and to correlate these findings with tile severity of the concurrent clinical and laboratory abnormalities. Material and methods
The study population consisted of obstetric patients managed at the E.H. Crump Women's Hospital of the Regional Medical Center, Memphis, Tennessee, and at Central Baptist Hospital, Lexington, Kentucky. Patients with the diagnosis of HELLP syndrome with complaints of severe right u p p e r quadrant abdominal pain in association with either shoulder or neck pain underwent imaging of the liver. Further, patients with HELLP syndrome and evidence of massive ascites, respiratory difficulty, or relapsing hypotension were evaluated to assess for the presence of intraparenchymal hemorrhage or subcapsular hematoma of the liven
Volmne 174, Number 6 AmJ ObstetGynecol
Barton and Sibai
1821
Table I. Serious maternal complications
Complication
} I
Severe ascites Pleural effusions Disseminated intravascular coagulopathy Acute renal failure Pulmonary edema Rupture of subcapsular hematoma of the liver Eclampsia
The primary imaging modality was computed tomography (CT), although in recent years we have increased the use of magnetic resonance imaging (MRI). The CT scans were performed on a Siemens (Erlangen, Ger' many) DR3 (third-generation) or a Siemens CRX 512 (third-generation) scanner. The MR/ unit used was a Siemens 1.0 T superconducting magnet. T1- and T2weighted spin echo sequences were used for MRI. A subcapsular hematoma was defined as a collection of hemorrhage between the liver capsule and the underlying liver parenchyma. Intraparenchymal hemorrhage was defined as a focus of hemorrhage within the liver parenchyma that was separated from the subcapsular space by an area of liver parenchyma. Liver infarction was considered tO be present if gas was noted within the liver parenchyma and there was no strong clinical evidence for infection. The diagnosis of HELLP syndrome was based on the following laboratory parameters as previously described by SibaiS: platelet count <100 x 109/L, serum aspartate aminotransferase >70 U/L, serum lactic dehydrogenase >600 U/L, or total bilirubin >1.2 mg/dl. Routine laboratory evaluation included serial measurements of liver function tests, complete blood cell count, coagulation profile, and renal function tests; Disseminated intravascular coagulation was defined as the presence of low platelets (<100 x 109/L), low fibrinogen (<300 mg/dl), positive fibrin split products (>40 l~g/dl), and prolonged prothrombin (>14 seconds) and partial thromboplastin times (>40 seconds). Acute renal failure was diagnosed in the presence of oliguria-anuria in association with severe reduction in renal function (creatinine clearance _<20 m l / m i n ) . The diagnosis of puhnonary edema was made on the basis of clinical and chest x-ray findings. Severe ascites was defined as the presence of large-volume ascites at either cesarean section or laparotomy, 9 or if noted on ultrasonography or CT of the abdomen. The results of this study are expressed as median (minimum, maximum) unless otherwise noted. The MannWhitney Urank test and Fisher's exact test were used for statistical analysis. A p value of <0.05 was considered significant.
Results A total of 34 patients were entered into the study. The clinical findings at the diagnosis of HELLP syndrome
No.
%
21 18 18 14 9 4 3
62 53 53 41 26 12 9
included a median maternal age of 24 years (range 16 to 40 years), median gestational age of 34 weeks (range 18 to 40 weeks), median systolic blood pressure 157 mm Hg (range 90 to 200 m m Hg), and median diastolic blood pressure of 100 mm Hg (range 50 to 130 m m Hg). Nineteen patients were white and 15 were black. Twenty-one patients were nulliparous, and 60% underwent cesarean delivery. Serious maternal complications occurring in these patients are presented in Table I. Twenty-nine patients required transfusion of blood or blood products. The median platelet transfusion was 10 units (range 0 to 90 units) and the median packed red blood cell transfusion was 5 units (range 0 to 58 units). There were 9 perinatal deaths and 1 maternal death. Sixteen patients had abnormal hepatic imaging results. Clinical characteristics and laboratory values for these patients are presented in Table II. The most common CT abnormalities were subcapsular hematoma of the liver ( n = 13) and intraparenchymal hemorrhage ( n = 6). A CT axial image of an infarction of the left lobe of the liver is presented in Fig. 1. MRI coronal and axial images of an u n r u p t u r e d subcapsular hematoma of the liver are depicted in Figs. 2 and 3, respectively. Comparison of the clinical characteristics and laboratory evaluations of patients with normal and abnormal hepatic imaging findings is presented in Table III. Statistical analysis demonstrated a significant difference in platelet count nadir between the patients with n o r m a l and abnormal imaging findings (p = 0.04) but failed to show any statistically significant difference in gestational age, mean arterial pressure, or the other laboratory parameters studied. Of the 13 patients with severe thrombocytopenia (platelet count <20 x 109/L), 10 (77%) had abnormal hepatic imaging findings. A separate statistical analysis for patients with and without a subcapsular hematoma of the liver failed to demonstrate any statistical difference for gestational age, mean arterial pressure, or the other laboratory parameters studied.
Comment Our goal of hepatic imaging was to define abnormalities that may be treated to help decrease the morbidity and mortality associated with HELLP syndrome. A second goal was to learn more about the pathophysiologic characteristics of the disease to help prevent or decrease
1822
Barton and Sibai
June 1996 AmJ Obstet Gynecol
Table II. Clinical characteristics and laboratory values in patients with HELLP syndrome and a b n o r m a l hepatic imaging findings
Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
I,atonaaeatI
xiouol
delivery (wk)
(ram H~
(xlOg/L)
AST (U/L)
LDH (U/L)
31 39 36 31 37 35 21 29 34 25 18 31 30 39 38 23
140 107 113 107 63 89 140 119 120 111 97 96 127 117 118 130
24 10 15 20 19 '48 7 13 12 8 40 9 48 52 54 17
4,026 879 3,670 6,193 774 74 178 2,332 9,297 365 5,340 289 277 849 7,248 540
2,331 2,000 7,600 23,584 1,084 1,035 1,320 2,893 39,411 7,580 1,500 1,780 1,460 1,068 27,08~ 2,013
I
bilirubin(mg/dl)
findings
16.2 10.6 20.0 1.9 5.0 3.2 2.9 3.6 10.0 4.8 3.4 6.8 2.4 10.6 1.7 2.1
* t § t * *,$ * * * * * t,$ *,$ t,;
MAP, Mean arterial blood pressure; AST, aspartate aminotransferase; LDH, lactate dehydrogenase. *Subcapsular hematoma of the liver, intact capsule. j-Subcapsular hematoma of the liver, ruptured capsule. $Intraparencbymal hemorrhage. §Liver infarction.
Table IlL C o m p a r i s o n of clinical characteristics and laboratory evaluations with hepatic imaging in H E L L P syndrome
Hepagcima~ng Normal (n = 18) History and physical examination Gestational age at delivery (wk) Mean arterial blood pressure (mm Hg) Laboratory findings Maximum uric acid (mg/dl) Maximum total bilirubin (mg/dl) Maximum lactate dehydrogenase (U/L) Maximum aspartate aminotransferase (U/L) Platelet count nadir (xl09/L)
I
Abnormal (n = 16)
Significance
34.5 (29, 40) 123 (90, 149)
31.0 (18, 39) 115 (63, 140)
p= 0.218 p = 0.195
8.55 (4.5, 16.1) 412 (1.0, 17.0)
9.05 (4.1, 20.6) 4.2 (1.7, 20.0)
p = 0.691 p = 0.743
2290 (713, 12864)
2006 (1035, 39,411)
p = 0.448
625 (121, 4020)
864 (74, 9297)
p = 0.285
30.0 (18, 86)
18.0 (7, 54)
p = 0.040
Values are median, with minimum and maximum in parentheses. Statistical analysis is by Mann-Whitney Urank test.
its o c c u r r e n c e in o t h e r women. Hepatic imaging abnormalities were n o t e d in 16 patients (47%). E m e r g e n c y i n t e r v e n t i o n was n e e d e d for 6 patients on the basis of these imaging findings. CT and MRI have excellent sensitivity for detecting acute liver h e m o r r h a g e , but because CT was m o r e available and faster and safer for potentially unstable patients it was the imaging modality of choice, Because o f the availability o f portable ultrasonography units, however, ultrasonography was used as a screening imaging modality. We recently c o m p a r e d hepatic histopathologic findings with clinical findings and the course of 11 patients with H E L L P syndrome. 5 T h e r e was no statistically signifi-
cant correlation between the severity- of the histotogic findings of periportal h e m o r r h a g e and fibrin deposition and the clinical laboratory findings. Similarly, in the curr e n t study we f o u n d no statistically significant correlation between the presence of an a b n o r m a l hepatic imaging finding and the severity of liver function test abnormalities. However, the severity of t h r o m b o c y t o p e n i a did correlate with hepatic imaging findings (p = 0.04). In particular, a platelet c o u n t of<20 x 10°/L was associated with an a b n o r m a l hepatic imaging finding for 10 of 13 pat e n t s (p = 0.012). T h e differential diagnosis of an u n r u p t u r e d subcapsular h e m a t o m a of the liver in pregnancy should include
Volume 174, Number 6 AmJ Obstet Gynecol
Barton and Sibai 1823
Fig. 1. CT axial image through liver and spleen. R, Maternal right; L, maternal left.
Fig. 2. Tl-weighted magnetic resonance coronal image through liver and spleen. acute fatty liver of pregnancy, abruptio placentae with disseminated intravascnlar coagulation, ruptured uterus, acute cholecystitis wit h sepsis, and thrombotic thrombocytopenia purpura. Most patients with a subcapsular hematoma of the liver are seen in the third or tate second trimester of pregnancy, although cases have been reported in the immediate postpartum period. I n addition to the signs and symptoms of preeclampsia,
physical examination findings consistent with peritoneal irritation and hepatomegaly may be present. Stimulation of the phrenic nerve at the diaphragm can produce referred pain along this nerve's distribution to its origin in the C4-C5 cervical plexus, including the pericardium, peritoneum, pleura, and shoulder. Because the gallbladder and esophagus share innervation by the phrenic nerve with the diaphragm, irritation of the
1824
Barton and Sibai
June 1996 .~nJ Obstet G;alecol
Fig. 3. Tl-weighted magnetic resonance axial image through liver and spleen.
I Unruptured-j
I Ruptured 1
I
Correct coagulopathy Close observation Serial ultrasound or computed tomography
I
V Hematoma size and Maternal condition stable
I Continued observation
_
Hematoma size increase and Maternal condition deteriorates
I Surgical evacuation
Surgical consultation Transfusions of blood and blood products Cross match and request adequate blood and blood products
I Laparotomy Evacuation and drainage of hematoma Packing if needed Suture laceration if possible Surgical mesh or omentum flap Embolization of hepatic artery
Fig. 4. Algorithm for management of subcapsular hematoma of liver in women with HELLP syndrome.
diaphragm may produce sensations of pain in these organs. Surgical repair has historically been r e c o m m e n d e d for hepatic hemorrhage without liver rupture. However, more recent experience suggests that this complication can be managed conservatively in patients who remain hemodynamically stable. 1° Management should include
close monitoring of hemodynamic and coagulation status. Serial assessment of the subcapsular hematoma with ultrasonography or CT is necessary if the clinical situation is not improving with immediate intervention for rupture or for deterioration of maternal status. O f extreme importance with conservative management is to avoid exogenous sources of trauma to the liver such as
Volume 174, Number 6 AmJ ObstetGynecol
abdominal palpation, convulsions, or emesis and to use care in transportation of the patient. Indeed, any sudden increase in intraabdorninat pressure could potentially lead to rupture of the subcapsular hematoma. 11 Rupture of a subcapsular hematoma of the liver is a life-threatening complication of HELLP syndrome. In most instances rupture involves the right lobe and is preceded by the development of a parenchymal hematoma. It usually presents with severe epigastric pain that may persist for several hours before circulatory collapse. 12 Patients frequently have shoulder pain, shock, evidence of massive ascites, respiratory difficulty, or pleural effusions. Often the fetus is dead at diagnosis. Four patients in this study had rupture of a subcapsular heinatoma of the liver; all required massive transfusions and laparotomy for control of hemorrhage. Three survived without long-term morbidity. On the basis of our experience and review of the literature, we have developed an algorithm for the management of hepatic complications of HELLP syndrome (Fig. 4). This algorithm emphasizes the potential for transfusion of large amounts of blood and blood products and the need for aggressive intervention if rupture of the hematoma is suspected. We recommend 30 units of packed red blood cells, 20 units of fresh-frozen plasma, 30 to 50 units of platelets, and 20 to 30 units of cryoprecipitate be available if rupture of a subcapsular hematoma is suspected. Our experience is in agreement with the recent observations of Smith et al.,7 in that a stable patient with an u n r u p t u r e d subcapsular hematoma should be conservatively managed. Constant monitoring must continue during this management, however, because patients can rapidly become unstable after rupture of the hematoma. Survival is dearly associated with rapid diagnosis and immediate medical or surgical stabilization. Coagulopathy must be aggressively managed because failure to do so is associated with an increased incidence of renal failure. In addition, these patients should be managed in an intensive care unit facilitywith close monitoring ofhemodynamic parameters and fluid status to avoid the potential for pulmonary edema or respiratory compromise. Postpartum follow-up for patients with subcapsular hematoma of the liver should include serial CT, MRI, or ultrasonography until the defect resolves. For patients receiving numerous transfusions, the hepatitis and human immunodeficiency virus status and isoantibody development should be assessed. Although the data on subsequent pregnancy outcome after a subcapsular hematoma of the liver in pregnancy are limited, we have managed two such patients who have had subsequent normal maternal and fetal outcomes.
REFERENCES
1. Weinstein L. Syndrome of hemolysis, elevated liver enzymes, and low platelet count: a severe consequence of hypertension in pregnancy. Am J Obstet Gynecol 1982;142:159-67. 2. Sibai BM, Taslimi M, E1-NazerA, Arnon E, Mabie BC, Ryan
Barton and Sibai
3.
4. 5.
6, 7, 8. 9.
10. 11. 12.
1825
GM. Maternal-perinatal outcome associated with the syndrome of hemolysis,elevated liver enzymes, and low platelets in severe preeclampsia-eclampsia. Am J Obstet Gynecol i986;155:501-9. AarnoudseJG, HouthoffHG, WeitsJ, VellengaE, Huisjes HJ. A syndrome of liver damage and intravascular coagulation in the last trimester of normotensive pregnancy: a clinical and histopathological study. BrJ Obstet Gynaecol 1986;93:14555. Arias F, Mancilla-Jimenez R. Hepatic fibrinogen deposits in preeclampsia--immunofluorescentevidence. N Engl J Med 1976;295:578-82. Barton JR, Riely CA, Adamec TA, Shanklin DR, Khoury A, Sibai BM. Hepatic histopathologic condition does not correlate with laboratory abnormalities in HELLP sydrome (hemolysis, elevated liver enzymes, and low platelet count). AmJ Obstet Gynecol 1992;167:1538-43. Goodlin RC, AndersonJC, Hodgson PE. Conservative treatment of liver hematoma in the postpartum period: a report of two cases.J Reprod Med 1985;30:368-70. SmithJGJr, Moise Kjj1; Dildy GA, Carpenter RJJr. Spontaneous rupture of the liver during pregnancy: current therapy. Obstet Gynecol 1991;77:171-5. Sibai BM. The HELLP syndrome (hemolysis, elevated liver enzymes, and low platelets) : much ado about nothing? AmJ Obstet Gynecol 1990;162:311-6. Woods JB, Blake PG, Perry KG Jr, Magann EF, Martin RW, Martin JN Jr. Ascites: a portent of cardiopuhnonary complications in the preeclamptic patient with the syndrome of hemolysis, elevated liver enzymes, and low platelets. Obstet Gynecol 1992;80:87-91. Manas KJ, WelshJD, Rankin RA, Miller DD. Hepatic hemorrhage without rupture in preeclampsia. N EnglJ Med 1985; 312:424-6. Neerhof MG, Zelman W, Sullivan T. Hepatic rupture in pregnancy: a review. Obstet Gynecol Surv 1989;44:407-9. Villegas H, Chavez-AzuelaJ, Calderon-Marquez ~, MaqueoTopete M. Spontaneous rupture of the liver in pregnancy toxemias: six cases and review of the world literature. Gineeol Obstet 1970;27:543-50.
Discussion
DR. M. MARKT ~ , Chattanooga, Tennessee. In this descriptive study the authors address one of the most dreaded complications of preeclampsia--hepatic hemorrhage. The rarity of hepatic hemorrhage in pregnancy may not allow a prospective randomized clinical trial to identify the best diagnostic means or treatment modality. That leaves us with retrospective chart reviews, which are plagued with serious problems in data accuracy and entirety. This study must be viewed and valued within those limitations. The b r u n t of this study is that abnormal liver function tests do not predict hepatic hemorrhage; platelet count does. The authors chose a nadir platelet count of 20 x 109/L as a discriminating cutoff value to predict the presence of abnormal hepatic imaging findings. Still, 5 of 24 (21%) patients with platelet counts above the cutoff value had hepatic hemorrhage, one of whom had a ruptured hematoma. In fact, of the four patients with ruptured liver hematomas, three had a nadir platelet count below, and one above, the cutoff value (Table II). Most published papers in our specialty contain statistical analyses. In search of statistical significance, the authors performed an appropriate nonparametric test on clinical and laboratory correlates of hepatic imaging abnormalities (Table III). Seven attributes were compared between the two subgroups with normal and abnormal hepatic imaging results. A p value <0.05 was predeter-
1826
June !996 ArnJ Obstet Gynecol
Barton and Sibai
mined to denote significance. The median of the lowest platelet count in each group was found to be significantly different. However, the nadir platelet count in both groups crossed over the previously mentioned cutoff value. There was only one maternal death among these 34 critically ill patients (Table I). These patients were indeed u n d e r close observation. The authors agree that surgical intervention should be reserved for patients with ruptured liver hematomas. Indeed, the algorithm requires surgery only when the maternal clinical condition deteriorates in spite of correction of coagulopathy. This recommendation makes perfect clinical sense. I would like to ask the authors to elaborate on the following questions. (1) Considering the retrospective design of the study and the multiplicity of the attributes compared, should you not apply a correction factor when calculating a posteriori probability? (2) Although close clinical observation and corrective management of coagulopathy is appropriately recommended for patients with u n r u p t u r e d liver hematomas and surgical intervention is reserved only for a few clinically deteriorating patients, why do the authors recommend serial hepatic imaging for all patients with u n r u p t u r e d subcapsular hematomas? (3) What is the inclusive period of the study? (4) Howwas gestational age established in the 18-week pregnant patient with HELLP syndrome? (5) Six patients had emergency intervention, four of whom had liver rupture. What complications and interventions did the other two have? (6) Do the authors recommend screening liver imaging for all patients with HELLP syndrome and shoulder pain? DR. CHRISTINECOMSTOCK,Royal Oaks Michigan. This report adds a large n u m b e r of cases tO what has been already published about imaging of hepatic hemorrhage in HELLP syndrome. It also tells us that CT evidence for hepatic hemorrhage is most likely to be found in those patients with the lowest platelet counts. These patients, out of all patients with HELLP syndrome, need to be observed particularly closely, even after delivery. Additionally, in cases in which the clinical findings are confusing and the diagnosis is not clear, CT and MRI can help distinguish among Budd-Chiari syndrome, fatty liver, infarction, and hemorrhage. However, in this era of cost containment and evidencebased medicine, we need to ask, once this relationship has b e e n established in the literature, how a CT scan would change clinical m a n a g e m e n t in the patient with classic HELLP syndrome. Barton et al. ~ have shown that the severity of liver abnormalities (assessed by biopsy) in HELLP syndrome could not be predicted by any clinical signs or laboratory findings and therefore advised aggressive management, that is, stabilization and delivery. If these patients will be delivered soon and will therefore on the road to recovery, how does it help us to know that they have intrahepatic hemorrhage? Could you explain why you suggest more than one scan in your algorithm? You did not do serial scans in your study, and. therefore I cannot determine how you arrived at the conclusion that they would be clinically useful. This paper confirms the findings of a smaller study by Manas et al. 2 in 1985. In that study all seven preeclamptic
patients with sudden-onset right upper quadrant pain had intraparenchymal or subcapsular hematomas by CT scan. Liver enzymes were not predictive, as you have also found. However, the hematomas were heralded by a sudden drop in hematocrit of an average of 10% and of platelets from >50,000/mm 3 to <50,000/ram 3. You have included one set of platelet and hematocrit values. When were those drawn in relation to the CT scans? Did you do serial platelet counts and hematocrits, and did you find a decrease in hematocrit also? Is there any way to predict who will have a ruptured liver? Did any patient with an u n r u p t n r e d intrahepatic hemorrhage proceed to a rupture and, if so, was the volume of intrahepatic blood seen on the CT scan predictive? In summary, the authors have described which patients may be at most risk for intrahepatic hemorrhage. Could the authors clarify how CT studies in patients with HELLP syndrome will affect management to improve Outcome? In other words, is hepatic CT a research or a clinical tool? REFERENCES
1. Barton JR, Riely CA, Adamec TA, Shanklin DR. Khoury AD, Sibai BM. Hepatic histopathologie condition does not correlate with laboratory abnormalities in HELLP syndrome (hemolysis, elevated liver enzymes, and low platelets). 3_'4J OBSTETG'ZX~COL1992;167:1538-43. 2. Manas KJ, WelshJD, Rankin RA, et al. Hepatic hemorrhage without rupture in preeclampsia. N EnglJ Med 1985;312:424. DP~WASH~GTONC. H l ~ , Sarasota, Florida. The message I get from this study is that maternal stabilization and delivery is still the way to treat HELLP syndrome and not to get a CT scan or an MRI. D~. PETERE I~HR, Minneapolis, Minnesota. Did any of these cases occur post partum? Did any ruptures occur in patients who had platelet counts <50,000/mm37 In other words, is there a critical platelet count at which it is more likely to occur? Did these patients come in with this severe pain or had they been watched for a while before the pain got that severe? In that case I would say the cure is delivery once the diagnosis of HELLP syndrome is made. My concern is whether these things occurred under Observation or did these patients come in late? In the one case that we've had with ruptured liver, the patient had been seen 3 days in a row with epigastric pain and had been ignored by the primary physician until she came in in shock with a ruptured liveI: DR. E~,NU~ P. G~z~2~o,Minneapolis, Minnesota. A few year ago Dr. Barton's group pointed out that the maternal mortality with HELLP syndrome was about 1 in 50. In Minnesota women get platelet counts and liver enzyme studies almost as often as they get blood pressure recordings. We generally effect delivery if the platelet count is <100,000 or near 100,000/mm ~. I wonder whether there were many of these delayed diagnoses of ruptured livers? D~ BARTOY(Closing). The study evaluated patients in Memphis from 1978 to 1995 and in Lexington from 1991 to 1995. Six patients had emergency interventions on the basis of imaging findings in the postpartum period. Four had ruptured subcapsular hematomas of the liver, which were confirmed at laparotomy. Three of the patients had
Volume 174, Number 6 AmJ Obstet Gynecol
a clinical diagnosis of abruptio placentae because of coagulopathy. After the fetuses were delivered vaginally, there wasn't the large a m o u n t of blood loss that we would normally expect with abruptio placentae. They began having complaints of right upper quadrant pain and pain in the shoulder. This is when the imaging was performed. These were not patients who had ruptured hematomas during conservative management before delivery. The fourth patient with a ruptured hematoma delivered vaginally, had a tubal ligation, and in the immediate postoperative period after her tubal ligation started complaining of severe shoulder pain. She had a ruptured hematoma of the liver at laparotomy. Of the final two patients who had emergency intervention, one patient had an intact subcapsular hematoma and a significant a m o u n t of intraperitoneal blood. We couldn't determine whether it was from a ruptured hematoma or from intraabdominal bleeding. At laparotomy she had generalized oozing that was believed to be caused by disseminated intravascular coagulation. We could find no active source of bleeding at laparotomy. The last patient had a laparotomy because of imaging findings suggestive of a hematoma and ascites. At laparotomy, multiple retroperitoneal hematomas were noted, probably related to the heparin therapy administered by our hematologist because he thought that she had disseminated intravascular coagulation. The gestational age for the patient who was seen at 18 weeks was established by first-trimester ultrasonography, and this was believed not to be a pregnancy complicated by severe intrauterine growth retardation. This patient had lost five previous pregnancies in the midtrimester from preeclampsia and two pregnancies after that in the midtrimester. She never achieved a live birth. She had an extensive evaluation for an antiphospholipid syndrome, which was negative. Concerning the need for which patient should have imaging of the liver, I believe this decision analysis is similar to the description we provided in our study on cerebral imaging in eclarnpsia published 3 years ago. 1Not all patients with eclampsia need to have a CT scan of the head. However, patients who have focal neurologic signs, coma, and those not improving in a reasonable amount of time and observation should be considered for cranial imaging. Similarly, not all patients with HELLP syndrome need liver imaging, but if the clinical situation is suspicious or complications such as intraparenchymal hemorrhage or subcapsular bleeding are suspected from irritation of the phrenic nerve and the patient is not improving as you would hope she would, then I believe hepatic imaging is indicated. Further, from our data we can't recommend routine imaging on the basis of severity of the liver functions but certainly would consider it in a patient with symptoms and a platelet count -<20 x 10~/L. As for our statistical analysis, I'll paraphrase one of the other presenters. I'm not a statistician, and I d o n ' t play one on television. We therefore consulted a statistician in
Barton a n d Sibai
1827
our study design and data analysis. There were only two groups compared, normal versus abnormal hepatic imaging; these were part of our study design from the onset. Observations were i n d e p e n d e n t measures, and therefore the Mann-Whitney Utest was chosen instead of a Wilcoxon rank sum test for statistical analysis. Further, most of the parameters we compared were necessary for confirming the diagnosis of HELLP syndrome (i.e., aspartate aminotransferase, lactate dehydrogenase, total bilirubin, and platelet count). There were only three variables that we compared in a posterior fashion: gestational age, mean arterial pressure, and uric acid, and none were statistically significant. For these three variables a posterior odds calculation would be appropriate and necessary before they could be considered statistically significant. As for Dr. Comstock's comments, I've already covered a few of them, but I would add that I believe imaging is a clinical, not a research, tool. Although the four patients with ruptured subcapsular hematomas had the rupture at imaging, no patient in our study had progression from an intact to a ruptured hematoma during management. Therefore I can't make any conclusions as to what would be predictive as far as such a volume of blood in the subcapsular hematoma and which patient might subsequently have a hematoma rupture. Platelet counts were presented as nadirs and reflect the severity of disease. They were obtained before imaging. There's an impression because of the trend in Memphis for using conservative management of severe preeclampsia that we're advocating conservative management of HELLP syndrome; nothing could be further from the truth. In their article on aggressive versus expectant management of severe preeclampsia at 28 to 32 weeks' gestation 2 and in their clinical commentary on conservative management of severe preeclampsia remote from term, 3 the authors from the Division of MaternalFetal Medicine at the University of Tennessee-Memphis identify HELLP syndrome as an indic~ttion for delivery. I would, however, advocate waiting 24 to 48 hours from the diagnosis of HELLP syndrome if the clinical situation is stable and the patient is remote from term to provide steroids for e n h a n c e m e n t of fetal lung maturity before proceeding with delivery. Certainly we do not believe patients with HELLP syndrome are candidates for prolonged conservative management as has been recently reported in the literature by several research groups in Europe using serial treatment with steroids. REFERENCES
1. Dahmus MA, Barton JR, Sibai BM." Cerebral imaging in eclampsia: magnetic resonance imaging versus computed tornography. AmJ Obstet Gynecol 1992;167:935-41. 2. Sibai BM, Mercer BM, Schiff E, Friedman SA. Aggressive versus expectant management of severe preeclampsia at 28 to 32 weeks' gestation: a randomized controlled trial. AmJ Obstet Gynecol 1994;171:818-22. 3. Schiff E, Friedman SA, Sibai BM. Conservative management of severe preeclampsia remote from term. Obstet Gynecol 1994;84:626-30.
OBJECTIVES: Our objective was to describe the hepatic imaging findings in selected patients with HELLP syndrome (hemolysis, elevated liver enzymes, and low platelet count) and to correlate these findings with the severity of concurrent clinical and laboratory abnormalities. STUDY DESIGN: Patients with laboratory criteria for HELLP syndrome with complaints of severe right upper quadrant abdominal pain in association with either shoulder pain, neck pain, or relapsing hypotension underwent imaging of the liver. Clinical and laboratory parameters were then correlated with the hepatic imaging findings. RESULTS: Thirty-four patients were evaluated in this study. Computed tomographic scanning of the liver was used for 33 patients. Additional imaging evaluations included magnetic resonance imaging for 4 patients and ultrasonographic evaluation of the liver for 5 patients. In 15 cases (45%) the computed tomographic results were abnormal. The most frequent abnormal hepatic imaging findings were subcapsular hematoma (n = 13) and intraparenchymal hemorrhage (n = 6). There was no statistically significant correlation between the presence of an abnormal hepatic imaging finding and the severity of liver function test abnormalities. However, the severity of thrombocytopenia did correlate with hepatic imaging findings (p = 0.04). In particular, an abnormal hepatic imaging finding was noted for 10 of 13 patients (77%) with a platelet count of <20 x 109/L (p = 0.012). CONCLUSIONS: Abnormalities in liver function test results do not accurately reflect the presence of abnormal hepatic imaging findings in HELLP syndrome. Patients with HELLP syndrome having complaints of right upper quadrant pain and neck pain, shoulder pain, or relapsing hypotension should undergo imaging of the liver. (Am J Obstet Gynecol 1996;174:1820-7.)
Key words: Preeclampsia, HELLP syndrome, computed tomography, magnetic resonance imaging
Hemolysis, abnormal liver function tests, and thrombocytopenia have been recognized as complications of preeclampsia-eclampsia for many years. In 1982 Weinstein 1 described 29 cases of severe preeclampsia-eclampsia in which the above complications occurred and coined the term HELLP syndrome, H f o r hemolysis, E L for elevated liver enzymes, and LP for low platelets. Patients with HELLP syndrome may have various signs and symptoms, although Sibai et al. 2 noted that 90% of patients with HELLP syndrome complain of epigastric or right upper quadrant pain. Right u p p e r quadrant or epigastric pain is thought to result from obstruction of blood flow in the hepatic sinusoids because of large hyaline deposits of fibrin-like material associated with periportal or focal pa-
From the Department of Obstetrics and Gynecology, Central Baptist Hospital, ~ and the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Tennessee, Memphis. b Annual Central Prize Award, presented at the Sixty-third Annual Meeting of The Central Association of Obstetricians and Gynecologists, Palm Desert, California, October 19-21, 1995. Reprint requests: John R. Barton, MD, Perinatal Diagnostic Cent~ Central Baptist Hospital, 1740 Nicholasville Road, Lexington, KY 40503. Copyright © 1996 by Mosby-Year Book, Inc. 0002-9378/96 $5.00+ 0 6/6/72246 1820
renchymal necrosis? -5 These histopathologic findings may be related to the elevated liver enzymes commonly observed in patients with this syndrome. In certain cases the cellular necrosis and hemorrhage may be severe enough to result in liver infarction or formation of a subcapsular hematoma. 6' 7 The purpose of this study was to describe the hepatic imaging findings in selected patients with HELLP syndrome and to correlate these findings with tile severity of the concurrent clinical and laboratory abnormalities. Material and methods
The study population consisted of obstetric patients managed at the E.H. Crump Women's Hospital of the Regional Medical Center, Memphis, Tennessee, and at Central Baptist Hospital, Lexington, Kentucky. Patients with the diagnosis of HELLP syndrome with complaints of severe right u p p e r quadrant abdominal pain in association with either shoulder or neck pain underwent imaging of the liver. Further, patients with HELLP syndrome and evidence of massive ascites, respiratory difficulty, or relapsing hypotension were evaluated to assess for the presence of intraparenchymal hemorrhage or subcapsular hematoma of the liven
Volmne 174, Number 6 AmJ ObstetGynecol
Barton and Sibai
1821
Table I. Serious maternal complications
Complication
} I
Severe ascites Pleural effusions Disseminated intravascular coagulopathy Acute renal failure Pulmonary edema Rupture of subcapsular hematoma of the liver Eclampsia
The primary imaging modality was computed tomography (CT), although in recent years we have increased the use of magnetic resonance imaging (MRI). The CT scans were performed on a Siemens (Erlangen, Ger' many) DR3 (third-generation) or a Siemens CRX 512 (third-generation) scanner. The MR/ unit used was a Siemens 1.0 T superconducting magnet. T1- and T2weighted spin echo sequences were used for MRI. A subcapsular hematoma was defined as a collection of hemorrhage between the liver capsule and the underlying liver parenchyma. Intraparenchymal hemorrhage was defined as a focus of hemorrhage within the liver parenchyma that was separated from the subcapsular space by an area of liver parenchyma. Liver infarction was considered tO be present if gas was noted within the liver parenchyma and there was no strong clinical evidence for infection. The diagnosis of HELLP syndrome was based on the following laboratory parameters as previously described by SibaiS: platelet count <100 x 109/L, serum aspartate aminotransferase >70 U/L, serum lactic dehydrogenase >600 U/L, or total bilirubin >1.2 mg/dl. Routine laboratory evaluation included serial measurements of liver function tests, complete blood cell count, coagulation profile, and renal function tests; Disseminated intravascular coagulation was defined as the presence of low platelets (<100 x 109/L), low fibrinogen (<300 mg/dl), positive fibrin split products (>40 l~g/dl), and prolonged prothrombin (>14 seconds) and partial thromboplastin times (>40 seconds). Acute renal failure was diagnosed in the presence of oliguria-anuria in association with severe reduction in renal function (creatinine clearance _<20 m l / m i n ) . The diagnosis of puhnonary edema was made on the basis of clinical and chest x-ray findings. Severe ascites was defined as the presence of large-volume ascites at either cesarean section or laparotomy, 9 or if noted on ultrasonography or CT of the abdomen. The results of this study are expressed as median (minimum, maximum) unless otherwise noted. The MannWhitney Urank test and Fisher's exact test were used for statistical analysis. A p value of <0.05 was considered significant.
Results A total of 34 patients were entered into the study. The clinical findings at the diagnosis of HELLP syndrome
No.
%
21 18 18 14 9 4 3
62 53 53 41 26 12 9
included a median maternal age of 24 years (range 16 to 40 years), median gestational age of 34 weeks (range 18 to 40 weeks), median systolic blood pressure 157 mm Hg (range 90 to 200 m m Hg), and median diastolic blood pressure of 100 mm Hg (range 50 to 130 m m Hg). Nineteen patients were white and 15 were black. Twenty-one patients were nulliparous, and 60% underwent cesarean delivery. Serious maternal complications occurring in these patients are presented in Table I. Twenty-nine patients required transfusion of blood or blood products. The median platelet transfusion was 10 units (range 0 to 90 units) and the median packed red blood cell transfusion was 5 units (range 0 to 58 units). There were 9 perinatal deaths and 1 maternal death. Sixteen patients had abnormal hepatic imaging results. Clinical characteristics and laboratory values for these patients are presented in Table II. The most common CT abnormalities were subcapsular hematoma of the liver ( n = 13) and intraparenchymal hemorrhage ( n = 6). A CT axial image of an infarction of the left lobe of the liver is presented in Fig. 1. MRI coronal and axial images of an u n r u p t u r e d subcapsular hematoma of the liver are depicted in Figs. 2 and 3, respectively. Comparison of the clinical characteristics and laboratory evaluations of patients with normal and abnormal hepatic imaging findings is presented in Table III. Statistical analysis demonstrated a significant difference in platelet count nadir between the patients with n o r m a l and abnormal imaging findings (p = 0.04) but failed to show any statistically significant difference in gestational age, mean arterial pressure, or the other laboratory parameters studied. Of the 13 patients with severe thrombocytopenia (platelet count <20 x 109/L), 10 (77%) had abnormal hepatic imaging findings. A separate statistical analysis for patients with and without a subcapsular hematoma of the liver failed to demonstrate any statistical difference for gestational age, mean arterial pressure, or the other laboratory parameters studied.
Comment Our goal of hepatic imaging was to define abnormalities that may be treated to help decrease the morbidity and mortality associated with HELLP syndrome. A second goal was to learn more about the pathophysiologic characteristics of the disease to help prevent or decrease
1822
Barton and Sibai
June 1996 AmJ Obstet Gynecol
Table II. Clinical characteristics and laboratory values in patients with HELLP syndrome and a b n o r m a l hepatic imaging findings
Patient No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
I,atonaaeatI
xiouol
delivery (wk)
(ram H~
(xlOg/L)
AST (U/L)
LDH (U/L)
31 39 36 31 37 35 21 29 34 25 18 31 30 39 38 23
140 107 113 107 63 89 140 119 120 111 97 96 127 117 118 130
24 10 15 20 19 '48 7 13 12 8 40 9 48 52 54 17
4,026 879 3,670 6,193 774 74 178 2,332 9,297 365 5,340 289 277 849 7,248 540
2,331 2,000 7,600 23,584 1,084 1,035 1,320 2,893 39,411 7,580 1,500 1,780 1,460 1,068 27,08~ 2,013
I
bilirubin(mg/dl)
findings
16.2 10.6 20.0 1.9 5.0 3.2 2.9 3.6 10.0 4.8 3.4 6.8 2.4 10.6 1.7 2.1
* t § t * *,$ * * * * * t,$ *,$ t,;
MAP, Mean arterial blood pressure; AST, aspartate aminotransferase; LDH, lactate dehydrogenase. *Subcapsular hematoma of the liver, intact capsule. j-Subcapsular hematoma of the liver, ruptured capsule. $Intraparencbymal hemorrhage. §Liver infarction.
Table IlL C o m p a r i s o n of clinical characteristics and laboratory evaluations with hepatic imaging in H E L L P syndrome
Hepagcima~ng Normal (n = 18) History and physical examination Gestational age at delivery (wk) Mean arterial blood pressure (mm Hg) Laboratory findings Maximum uric acid (mg/dl) Maximum total bilirubin (mg/dl) Maximum lactate dehydrogenase (U/L) Maximum aspartate aminotransferase (U/L) Platelet count nadir (xl09/L)
I
Abnormal (n = 16)
Significance
34.5 (29, 40) 123 (90, 149)
31.0 (18, 39) 115 (63, 140)
p= 0.218 p = 0.195
8.55 (4.5, 16.1) 412 (1.0, 17.0)
9.05 (4.1, 20.6) 4.2 (1.7, 20.0)
p = 0.691 p = 0.743
2290 (713, 12864)
2006 (1035, 39,411)
p = 0.448
625 (121, 4020)
864 (74, 9297)
p = 0.285
30.0 (18, 86)
18.0 (7, 54)
p = 0.040
Values are median, with minimum and maximum in parentheses. Statistical analysis is by Mann-Whitney Urank test.
its o c c u r r e n c e in o t h e r women. Hepatic imaging abnormalities were n o t e d in 16 patients (47%). E m e r g e n c y i n t e r v e n t i o n was n e e d e d for 6 patients on the basis of these imaging findings. CT and MRI have excellent sensitivity for detecting acute liver h e m o r r h a g e , but because CT was m o r e available and faster and safer for potentially unstable patients it was the imaging modality of choice, Because o f the availability o f portable ultrasonography units, however, ultrasonography was used as a screening imaging modality. We recently c o m p a r e d hepatic histopathologic findings with clinical findings and the course of 11 patients with H E L L P syndrome. 5 T h e r e was no statistically signifi-
cant correlation between the severity- of the histotogic findings of periportal h e m o r r h a g e and fibrin deposition and the clinical laboratory findings. Similarly, in the curr e n t study we f o u n d no statistically significant correlation between the presence of an a b n o r m a l hepatic imaging finding and the severity of liver function test abnormalities. However, the severity of t h r o m b o c y t o p e n i a did correlate with hepatic imaging findings (p = 0.04). In particular, a platelet c o u n t of<20 x 10°/L was associated with an a b n o r m a l hepatic imaging finding for 10 of 13 pat e n t s (p = 0.012). T h e differential diagnosis of an u n r u p t u r e d subcapsular h e m a t o m a of the liver in pregnancy should include
Volume 174, Number 6 AmJ Obstet Gynecol
Barton and Sibai 1823
Fig. 1. CT axial image through liver and spleen. R, Maternal right; L, maternal left.
Fig. 2. Tl-weighted magnetic resonance coronal image through liver and spleen. acute fatty liver of pregnancy, abruptio placentae with disseminated intravascnlar coagulation, ruptured uterus, acute cholecystitis wit h sepsis, and thrombotic thrombocytopenia purpura. Most patients with a subcapsular hematoma of the liver are seen in the third or tate second trimester of pregnancy, although cases have been reported in the immediate postpartum period. I n addition to the signs and symptoms of preeclampsia,
physical examination findings consistent with peritoneal irritation and hepatomegaly may be present. Stimulation of the phrenic nerve at the diaphragm can produce referred pain along this nerve's distribution to its origin in the C4-C5 cervical plexus, including the pericardium, peritoneum, pleura, and shoulder. Because the gallbladder and esophagus share innervation by the phrenic nerve with the diaphragm, irritation of the
1824
Barton and Sibai
June 1996 .~nJ Obstet G;alecol
Fig. 3. Tl-weighted magnetic resonance axial image through liver and spleen.
I Unruptured-j
I Ruptured 1
I
Correct coagulopathy Close observation Serial ultrasound or computed tomography
I
V Hematoma size and Maternal condition stable
I Continued observation
_
Hematoma size increase and Maternal condition deteriorates
I Surgical evacuation
Surgical consultation Transfusions of blood and blood products Cross match and request adequate blood and blood products
I Laparotomy Evacuation and drainage of hematoma Packing if needed Suture laceration if possible Surgical mesh or omentum flap Embolization of hepatic artery
Fig. 4. Algorithm for management of subcapsular hematoma of liver in women with HELLP syndrome.
diaphragm may produce sensations of pain in these organs. Surgical repair has historically been r e c o m m e n d e d for hepatic hemorrhage without liver rupture. However, more recent experience suggests that this complication can be managed conservatively in patients who remain hemodynamically stable. 1° Management should include
close monitoring of hemodynamic and coagulation status. Serial assessment of the subcapsular hematoma with ultrasonography or CT is necessary if the clinical situation is not improving with immediate intervention for rupture or for deterioration of maternal status. O f extreme importance with conservative management is to avoid exogenous sources of trauma to the liver such as
Volume 174, Number 6 AmJ ObstetGynecol
abdominal palpation, convulsions, or emesis and to use care in transportation of the patient. Indeed, any sudden increase in intraabdorninat pressure could potentially lead to rupture of the subcapsular hematoma. 11 Rupture of a subcapsular hematoma of the liver is a life-threatening complication of HELLP syndrome. In most instances rupture involves the right lobe and is preceded by the development of a parenchymal hematoma. It usually presents with severe epigastric pain that may persist for several hours before circulatory collapse. 12 Patients frequently have shoulder pain, shock, evidence of massive ascites, respiratory difficulty, or pleural effusions. Often the fetus is dead at diagnosis. Four patients in this study had rupture of a subcapsular heinatoma of the liver; all required massive transfusions and laparotomy for control of hemorrhage. Three survived without long-term morbidity. On the basis of our experience and review of the literature, we have developed an algorithm for the management of hepatic complications of HELLP syndrome (Fig. 4). This algorithm emphasizes the potential for transfusion of large amounts of blood and blood products and the need for aggressive intervention if rupture of the hematoma is suspected. We recommend 30 units of packed red blood cells, 20 units of fresh-frozen plasma, 30 to 50 units of platelets, and 20 to 30 units of cryoprecipitate be available if rupture of a subcapsular hematoma is suspected. Our experience is in agreement with the recent observations of Smith et al.,7 in that a stable patient with an u n r u p t u r e d subcapsular hematoma should be conservatively managed. Constant monitoring must continue during this management, however, because patients can rapidly become unstable after rupture of the hematoma. Survival is dearly associated with rapid diagnosis and immediate medical or surgical stabilization. Coagulopathy must be aggressively managed because failure to do so is associated with an increased incidence of renal failure. In addition, these patients should be managed in an intensive care unit facilitywith close monitoring ofhemodynamic parameters and fluid status to avoid the potential for pulmonary edema or respiratory compromise. Postpartum follow-up for patients with subcapsular hematoma of the liver should include serial CT, MRI, or ultrasonography until the defect resolves. For patients receiving numerous transfusions, the hepatitis and human immunodeficiency virus status and isoantibody development should be assessed. Although the data on subsequent pregnancy outcome after a subcapsular hematoma of the liver in pregnancy are limited, we have managed two such patients who have had subsequent normal maternal and fetal outcomes.
REFERENCES
1. Weinstein L. Syndrome of hemolysis, elevated liver enzymes, and low platelet count: a severe consequence of hypertension in pregnancy. Am J Obstet Gynecol 1982;142:159-67. 2. Sibai BM, Taslimi M, E1-NazerA, Arnon E, Mabie BC, Ryan
Barton and Sibai
3.
4. 5.
6, 7, 8. 9.
10. 11. 12.
1825
GM. Maternal-perinatal outcome associated with the syndrome of hemolysis,elevated liver enzymes, and low platelets in severe preeclampsia-eclampsia. Am J Obstet Gynecol i986;155:501-9. AarnoudseJG, HouthoffHG, WeitsJ, VellengaE, Huisjes HJ. A syndrome of liver damage and intravascular coagulation in the last trimester of normotensive pregnancy: a clinical and histopathological study. BrJ Obstet Gynaecol 1986;93:14555. Arias F, Mancilla-Jimenez R. Hepatic fibrinogen deposits in preeclampsia--immunofluorescentevidence. N Engl J Med 1976;295:578-82. Barton JR, Riely CA, Adamec TA, Shanklin DR, Khoury A, Sibai BM. Hepatic histopathologic condition does not correlate with laboratory abnormalities in HELLP sydrome (hemolysis, elevated liver enzymes, and low platelet count). AmJ Obstet Gynecol 1992;167:1538-43. Goodlin RC, AndersonJC, Hodgson PE. Conservative treatment of liver hematoma in the postpartum period: a report of two cases.J Reprod Med 1985;30:368-70. SmithJGJr, Moise Kjj1; Dildy GA, Carpenter RJJr. Spontaneous rupture of the liver during pregnancy: current therapy. Obstet Gynecol 1991;77:171-5. Sibai BM. The HELLP syndrome (hemolysis, elevated liver enzymes, and low platelets) : much ado about nothing? AmJ Obstet Gynecol 1990;162:311-6. Woods JB, Blake PG, Perry KG Jr, Magann EF, Martin RW, Martin JN Jr. Ascites: a portent of cardiopuhnonary complications in the preeclamptic patient with the syndrome of hemolysis, elevated liver enzymes, and low platelets. Obstet Gynecol 1992;80:87-91. Manas KJ, WelshJD, Rankin RA, Miller DD. Hepatic hemorrhage without rupture in preeclampsia. N EnglJ Med 1985; 312:424-6. Neerhof MG, Zelman W, Sullivan T. Hepatic rupture in pregnancy: a review. Obstet Gynecol Surv 1989;44:407-9. Villegas H, Chavez-AzuelaJ, Calderon-Marquez ~, MaqueoTopete M. Spontaneous rupture of the liver in pregnancy toxemias: six cases and review of the world literature. Gineeol Obstet 1970;27:543-50.
Discussion
DR. M. MARKT ~ , Chattanooga, Tennessee. In this descriptive study the authors address one of the most dreaded complications of preeclampsia--hepatic hemorrhage. The rarity of hepatic hemorrhage in pregnancy may not allow a prospective randomized clinical trial to identify the best diagnostic means or treatment modality. That leaves us with retrospective chart reviews, which are plagued with serious problems in data accuracy and entirety. This study must be viewed and valued within those limitations. The b r u n t of this study is that abnormal liver function tests do not predict hepatic hemorrhage; platelet count does. The authors chose a nadir platelet count of 20 x 109/L as a discriminating cutoff value to predict the presence of abnormal hepatic imaging findings. Still, 5 of 24 (21%) patients with platelet counts above the cutoff value had hepatic hemorrhage, one of whom had a ruptured hematoma. In fact, of the four patients with ruptured liver hematomas, three had a nadir platelet count below, and one above, the cutoff value (Table II). Most published papers in our specialty contain statistical analyses. In search of statistical significance, the authors performed an appropriate nonparametric test on clinical and laboratory correlates of hepatic imaging abnormalities (Table III). Seven attributes were compared between the two subgroups with normal and abnormal hepatic imaging results. A p value <0.05 was predeter-
1826
June !996 ArnJ Obstet Gynecol
Barton and Sibai
mined to denote significance. The median of the lowest platelet count in each group was found to be significantly different. However, the nadir platelet count in both groups crossed over the previously mentioned cutoff value. There was only one maternal death among these 34 critically ill patients (Table I). These patients were indeed u n d e r close observation. The authors agree that surgical intervention should be reserved for patients with ruptured liver hematomas. Indeed, the algorithm requires surgery only when the maternal clinical condition deteriorates in spite of correction of coagulopathy. This recommendation makes perfect clinical sense. I would like to ask the authors to elaborate on the following questions. (1) Considering the retrospective design of the study and the multiplicity of the attributes compared, should you not apply a correction factor when calculating a posteriori probability? (2) Although close clinical observation and corrective management of coagulopathy is appropriately recommended for patients with u n r u p t u r e d liver hematomas and surgical intervention is reserved only for a few clinically deteriorating patients, why do the authors recommend serial hepatic imaging for all patients with u n r u p t u r e d subcapsular hematomas? (3) What is the inclusive period of the study? (4) Howwas gestational age established in the 18-week pregnant patient with HELLP syndrome? (5) Six patients had emergency intervention, four of whom had liver rupture. What complications and interventions did the other two have? (6) Do the authors recommend screening liver imaging for all patients with HELLP syndrome and shoulder pain? DR. CHRISTINECOMSTOCK,Royal Oaks Michigan. This report adds a large n u m b e r of cases tO what has been already published about imaging of hepatic hemorrhage in HELLP syndrome. It also tells us that CT evidence for hepatic hemorrhage is most likely to be found in those patients with the lowest platelet counts. These patients, out of all patients with HELLP syndrome, need to be observed particularly closely, even after delivery. Additionally, in cases in which the clinical findings are confusing and the diagnosis is not clear, CT and MRI can help distinguish among Budd-Chiari syndrome, fatty liver, infarction, and hemorrhage. However, in this era of cost containment and evidencebased medicine, we need to ask, once this relationship has b e e n established in the literature, how a CT scan would change clinical m a n a g e m e n t in the patient with classic HELLP syndrome. Barton et al. ~ have shown that the severity of liver abnormalities (assessed by biopsy) in HELLP syndrome could not be predicted by any clinical signs or laboratory findings and therefore advised aggressive management, that is, stabilization and delivery. If these patients will be delivered soon and will therefore on the road to recovery, how does it help us to know that they have intrahepatic hemorrhage? Could you explain why you suggest more than one scan in your algorithm? You did not do serial scans in your study, and. therefore I cannot determine how you arrived at the conclusion that they would be clinically useful. This paper confirms the findings of a smaller study by Manas et al. 2 in 1985. In that study all seven preeclamptic
patients with sudden-onset right upper quadrant pain had intraparenchymal or subcapsular hematomas by CT scan. Liver enzymes were not predictive, as you have also found. However, the hematomas were heralded by a sudden drop in hematocrit of an average of 10% and of platelets from >50,000/mm 3 to <50,000/ram 3. You have included one set of platelet and hematocrit values. When were those drawn in relation to the CT scans? Did you do serial platelet counts and hematocrits, and did you find a decrease in hematocrit also? Is there any way to predict who will have a ruptured liver? Did any patient with an u n r u p t n r e d intrahepatic hemorrhage proceed to a rupture and, if so, was the volume of intrahepatic blood seen on the CT scan predictive? In summary, the authors have described which patients may be at most risk for intrahepatic hemorrhage. Could the authors clarify how CT studies in patients with HELLP syndrome will affect management to improve Outcome? In other words, is hepatic CT a research or a clinical tool? REFERENCES
1. Barton JR, Riely CA, Adamec TA, Shanklin DR. Khoury AD, Sibai BM. Hepatic histopathologie condition does not correlate with laboratory abnormalities in HELLP syndrome (hemolysis, elevated liver enzymes, and low platelets). 3_'4J OBSTETG'ZX~COL1992;167:1538-43. 2. Manas KJ, WelshJD, Rankin RA, et al. Hepatic hemorrhage without rupture in preeclampsia. N EnglJ Med 1985;312:424. DP~WASH~GTONC. H l ~ , Sarasota, Florida. The message I get from this study is that maternal stabilization and delivery is still the way to treat HELLP syndrome and not to get a CT scan or an MRI. D~. PETERE I~HR, Minneapolis, Minnesota. Did any of these cases occur post partum? Did any ruptures occur in patients who had platelet counts <50,000/mm37 In other words, is there a critical platelet count at which it is more likely to occur? Did these patients come in with this severe pain or had they been watched for a while before the pain got that severe? In that case I would say the cure is delivery once the diagnosis of HELLP syndrome is made. My concern is whether these things occurred under Observation or did these patients come in late? In the one case that we've had with ruptured liver, the patient had been seen 3 days in a row with epigastric pain and had been ignored by the primary physician until she came in in shock with a ruptured liveI: DR. E~,NU~ P. G~z~2~o,Minneapolis, Minnesota. A few year ago Dr. Barton's group pointed out that the maternal mortality with HELLP syndrome was about 1 in 50. In Minnesota women get platelet counts and liver enzyme studies almost as often as they get blood pressure recordings. We generally effect delivery if the platelet count is <100,000 or near 100,000/mm ~. I wonder whether there were many of these delayed diagnoses of ruptured livers? D~ BARTOY(Closing). The study evaluated patients in Memphis from 1978 to 1995 and in Lexington from 1991 to 1995. Six patients had emergency interventions on the basis of imaging findings in the postpartum period. Four had ruptured subcapsular hematomas of the liver, which were confirmed at laparotomy. Three of the patients had
Volume 174, Number 6 AmJ Obstet Gynecol
a clinical diagnosis of abruptio placentae because of coagulopathy. After the fetuses were delivered vaginally, there wasn't the large a m o u n t of blood loss that we would normally expect with abruptio placentae. They began having complaints of right upper quadrant pain and pain in the shoulder. This is when the imaging was performed. These were not patients who had ruptured hematomas during conservative management before delivery. The fourth patient with a ruptured hematoma delivered vaginally, had a tubal ligation, and in the immediate postoperative period after her tubal ligation started complaining of severe shoulder pain. She had a ruptured hematoma of the liver at laparotomy. Of the final two patients who had emergency intervention, one patient had an intact subcapsular hematoma and a significant a m o u n t of intraperitoneal blood. We couldn't determine whether it was from a ruptured hematoma or from intraabdominal bleeding. At laparotomy she had generalized oozing that was believed to be caused by disseminated intravascular coagulation. We could find no active source of bleeding at laparotomy. The last patient had a laparotomy because of imaging findings suggestive of a hematoma and ascites. At laparotomy, multiple retroperitoneal hematomas were noted, probably related to the heparin therapy administered by our hematologist because he thought that she had disseminated intravascular coagulation. The gestational age for the patient who was seen at 18 weeks was established by first-trimester ultrasonography, and this was believed not to be a pregnancy complicated by severe intrauterine growth retardation. This patient had lost five previous pregnancies in the midtrimester from preeclampsia and two pregnancies after that in the midtrimester. She never achieved a live birth. She had an extensive evaluation for an antiphospholipid syndrome, which was negative. Concerning the need for which patient should have imaging of the liver, I believe this decision analysis is similar to the description we provided in our study on cerebral imaging in eclarnpsia published 3 years ago. 1Not all patients with eclampsia need to have a CT scan of the head. However, patients who have focal neurologic signs, coma, and those not improving in a reasonable amount of time and observation should be considered for cranial imaging. Similarly, not all patients with HELLP syndrome need liver imaging, but if the clinical situation is suspicious or complications such as intraparenchymal hemorrhage or subcapsular bleeding are suspected from irritation of the phrenic nerve and the patient is not improving as you would hope she would, then I believe hepatic imaging is indicated. Further, from our data we can't recommend routine imaging on the basis of severity of the liver functions but certainly would consider it in a patient with symptoms and a platelet count -<20 x 10~/L. As for our statistical analysis, I'll paraphrase one of the other presenters. I'm not a statistician, and I d o n ' t play one on television. We therefore consulted a statistician in
Barton a n d Sibai
1827
our study design and data analysis. There were only two groups compared, normal versus abnormal hepatic imaging; these were part of our study design from the onset. Observations were i n d e p e n d e n t measures, and therefore the Mann-Whitney Utest was chosen instead of a Wilcoxon rank sum test for statistical analysis. Further, most of the parameters we compared were necessary for confirming the diagnosis of HELLP syndrome (i.e., aspartate aminotransferase, lactate dehydrogenase, total bilirubin, and platelet count). There were only three variables that we compared in a posterior fashion: gestational age, mean arterial pressure, and uric acid, and none were statistically significant. For these three variables a posterior odds calculation would be appropriate and necessary before they could be considered statistically significant. As for Dr. Comstock's comments, I've already covered a few of them, but I would add that I believe imaging is a clinical, not a research, tool. Although the four patients with ruptured subcapsular hematomas had the rupture at imaging, no patient in our study had progression from an intact to a ruptured hematoma during management. Therefore I can't make any conclusions as to what would be predictive as far as such a volume of blood in the subcapsular hematoma and which patient might subsequently have a hematoma rupture. Platelet counts were presented as nadirs and reflect the severity of disease. They were obtained before imaging. There's an impression because of the trend in Memphis for using conservative management of severe preeclampsia that we're advocating conservative management of HELLP syndrome; nothing could be further from the truth. In their article on aggressive versus expectant management of severe preeclampsia at 28 to 32 weeks' gestation 2 and in their clinical commentary on conservative management of severe preeclampsia remote from term, 3 the authors from the Division of MaternalFetal Medicine at the University of Tennessee-Memphis identify HELLP syndrome as an indic~ttion for delivery. I would, however, advocate waiting 24 to 48 hours from the diagnosis of HELLP syndrome if the clinical situation is stable and the patient is remote from term to provide steroids for e n h a n c e m e n t of fetal lung maturity before proceeding with delivery. Certainly we do not believe patients with HELLP syndrome are candidates for prolonged conservative management as has been recently reported in the literature by several research groups in Europe using serial treatment with steroids. REFERENCES
1. Dahmus MA, Barton JR, Sibai BM." Cerebral imaging in eclampsia: magnetic resonance imaging versus computed tornography. AmJ Obstet Gynecol 1992;167:935-41. 2. Sibai BM, Mercer BM, Schiff E, Friedman SA. Aggressive versus expectant management of severe preeclampsia at 28 to 32 weeks' gestation: a randomized controlled trial. AmJ Obstet Gynecol 1994;171:818-22. 3. Schiff E, Friedman SA, Sibai BM. Conservative management of severe preeclampsia remote from term. Obstet Gynecol 1994;84:626-30.