The Utility of Evaluating Low Serum Albumin Gradient Ascites in Patients With Cirrhosis

Transcription

1 nature publishing group ORIGINAL CONTRIBUTIONS 1401 The Utility of Evaluating Low Serum Albumin Gradient Ascites in Patients With Cirrhosis Hashim E. K handw a l l a, M D 1, Ye m i Fas a k i n, M D 1 and Hashem B. E l - S e r ag, M D, MPH 1 OBJECTIVES: METHODS: Serum-ascites albumin gradient (SAAG) has been used extensively in the diagnostic wkup of patients with ascites. A SAAG level of < 1.1 g / dl is usually thought of as a result of nonptal hypertension etiologies, including malignancies, tuberculous peritonitis, and nephrotic syndrome. However, the predictive value of a low SAAG in patients with existing cirrhosis in whom the pretest probability of ptal hypertension is high is not clear. We identified all patients with a SAAG of < 1.1 g / dl during a 5-year period at a single large veterans affairs medical center. Cirrhosis was defined by clinical, histological, and radiological features. Nonptal hypertension causes of low SAAG were identified, including bacterial peritonitis, peritoneal carcinomatosis, nephrogenous ascites, tuberculous peritonitis, chylous ascites, and pancreatic ascites. RESULTS: We identified 92 patients (76 with cirrhosis and 16 with no cirrhosis) with ascites and a SAAG of < 1.1 g / dl. Of the 76 patients with cirrhosis, only 29 (38 % ) had an identifiable cause, most commonly primary bacterial peritonitis (11, 38 % ), followed by peritoneal carcinomatosis malignant ascites (8, 28 % ) and nephrotic syndrome (5, 17 % ). There were 47 patients with cirrhosis and a low SAAG f whom no etiology was identified. Thirty-three patients underwent a repeat paracentesis, 24 (73 % ) of whom changed to a high SAAG. On the other hand, the 16 patients with no cirrhosis had significantly lower SAAG (0.66 vs. 0.81), and most (12, 75 % ) had an identifiable cause of ascites. CONCLUSIONS: Evaluation of a SAAG < 1.1 g / dl in patients with known cirrhosis has low yield and is less likely to be helpful than that in patients without cirrhosis. A repeat paracentesis as part of the wkup is recommended. Further studies of low SAAG cutoffs are needed. Am J Gastroenterol 2009; 104: ; doi: /ajg ; published online 21 April 2009 INTRODUCTION Serum-ascites albumin gradient (SAAG) is extensively used in the diagnostic wkup of ascites (1 6). Runyon et al. showed high sensitivity (96.7 % ) f a high SAAG (defined as SAAG 1.1 g / dl) in diagnosing ascites related to ptal hypertension (1). In the same study, the etiologies of low SAAG (defined as SAAG < 1.1 g / dl) in patients without cirrhosis included malignant ascites, tuberculous peritonitis, and nephrotic syndrome. However, the utility of evaluating a low SAAG in patients with known cirrhosis and ptal hypertension has not been adequately established. These patients have been described as having mixed ascites, with the presence of ptal hypertension and a variable SAAG level secondary to another underlying diagnosis (7). In the study conducted by Runyon et al., on ly 42 (4.7 % ) of 901 ascites samples obtained from patients with cirrhosis had mixed ascites. The yield of further wkup to uncover other causes of a low SAAG, including tuberculous peritonitis, peritoneal carcinomatosis, and nephrotic syndrome in patients with cirrhosis, is not clear. We conducted this study to determine the utility of a diagnostic wkup f patients with cirrhosis and a low SAAG and to establish a secondary cause of ascites. We hypothesized that, 1 Sections of Gastroenterology and Health Services Research at the Michael E. DeBakey Veterans Affairs Medical Center and Bayl College of Medicine, Houston, Texas, USA. Crespondence: Hashem B. El-Serag, MD, MPH, Bayl College of Medicine, Medicine, The Michael E. DeBakey Veterans Affairs Medical Center, 2002 Holcombe Boulevard, Houston, Texas 77030, USA. hasheme@bcm.tmc.edu Received 29 October 2008; accepted 12 January 2009

2 1402 Khandwalla et al. in the presence of a high pretest likelihood f ptal hypertension, a low SAAG will have a low predictive value f secondary causes of ascites. METHODS We screened 405 patients at the Michael E. DeBakey Veterans Affairs Medical Center in Houston, TX, who had undergone paracentesis with measurement of both serum and ascites albumin levels between January 2000 and December We identified patients with at least one documented SAAG of < 1.1 g / dl ( n = 92). We conducted systematic abstraction of medical recds to collect infmation on several demographic features, clinical, labaty, pathologic, and radiographic data. We evaluated patients in this study on the basis of the presence of cirrhosis. Cirrhosis was defined as being present on the basis of radiological, pathological, other evidence of the presence of a nodular liver ptal hypertension. On the basis of specific diagnostic criteria ( Table 1 ), we determined whether these patients had an appropriate wkup to establish a secondary cause of a low SAAG. The SAAG results of a repeat paracentesis, if perfmed, were evaluated f patients who initially had an unknown etiology of a low SAAG. We searched f commonly described etiologies f low SAAG levels through ascites fluid analysis f neutrophil count; bacterial and fungal culture; acid-fast bacillus (AFB) culture; fluid cytology and imaging modalities, including ultrasonography, computerized tomography (CT), and magnetic resonance imaging (MRI), to rule out malignancies, peritoneal biopsy where indicated, and urine protein measurement. These etiologies include spontaneous bacterial peritonitis (SBP), secondary bacterial peritonitis, tuberculous peritonitis, nephrogenous ascites, ascites secondary to nephrotic syndrome, malignant ascites peritoneal carcinomatosis, and pancreatic ascites. The diagnosis of SBP was based on bacterial growth in an ascitic fluid collected in blood culture bottles inoculated at the bedside an ascitic absolute neutrophil count of at least cells / l (8). A diagnosis of secondary bacterial peritonitis was made if a perfated viscus, intestinal ischemia, and / obstruction were documented by a radiographic surgical evaluation. The diagnosis of mycobacterial peritonitis required mycobacterial growth in an ascitic fluid culture peritoneal biopsy positive results of a polymerase chain reaction scan of ascitic fluid f a mycobacterium. Mycobacterial peritonitis diagnosis was also made if a mycobacterial smear was positive in the presence of a positive purified protein derivative if ascitic fluid adenosine deaminase level was at least 40 IU / l (9). Nephrogenous ascites was diagnosed in dialysis-related ascites, and ascites resulting from nephrotic syndrome was diagnosed in the presence of 24 h urine collection f protein of at least 3.5 g / day (10) spot urine protein-to-creatinine ratio >3 (11). The diagnosis of Table 1. Diagnostic criteria employed in the study f secondary causes of SAAG < 1.1 g/dl Condition peritonitis Candida peritonitis Nephrotic syndrome Tuberculous peritonitis Pancreatic ascites Chylous ascites Secondary bacterial peritonitis Malignant ascites Diagnosis Ascites fl uid ANC > cells/l Bacterial growth in ascites fl uid Positive ascitic fl uid culture f Candida species 24-h urine protein >3.5 g Spot urine protein/creatinine ratio >3 Positive ascites fl uid AFB smear with positive PPD Positive ascites fl uid culture f MTB Positive PPD and ascites fl uid adenosine deaminase > 40 IU/l Granulomatous infl ammation on peritoneal biopsy Positive ascites fl uid MTB PCR Ascites fl uid amylase level at least twice upper limit of nmal Ascites fl uid triglycerides >200 mg/dl Perfated viscus, ischemia, obstruction on radiographic study Positive ascites fl uid cytology Positive peritoneal biopsy biopsy of gan involved Tum localization as determined by radiographic study AFB, acid-fast bacillus; ANC, absolute neutrophil count; MTB, Mycobacterium tuberculosis ; PCR, polymerase chain reaction; PPD, purifi ed protein derivative; SAAG, serum-ascites albumin gradient. malignant ascites was based on positive cytology of an ascitic fluid (12), positive biopsy of the gan involved, tum localization as determined by CT scan, MRI, laparotomy, autopsy. Pancreatic ascites was diagnosed when the ascitic fluid amylase level was at least twice that of the upper limit of nmal f serum amylase (13). Lastly, the diagnosis of chylous ascites was based on ascitic fluid triglyceride level of at least 200 mg / dl (14). We used standardized comprehensive data collection fms designed in Telefm (Autonomy Cardiff, Vista, CA), which we subsequently scanned into an Access database and converted into SAS (Statistical Analysis Systems, Cary, NC) data sets f statistical analyses. Descriptive statistics were generated: mean and s.d. f continuous variables and proptions f categical variables. We compared patients with and without cirrhosis using chi-square f categical variables and t -test f continuous variables. The American Journal of GASTROENTEROLOGY VOLUME 104 JUNE

3 Utility of Serum Albumin Gradient Ascites 1403 RESULTS Of the 405 patients with ascites, 92 (23 % ) had at least one SAAG level of < 1.1 g / dl, of whom 76 had cirrhosis and 16 did not. There were 83 (90 % ) male patients. The mean age was 59 (s.d. = 10). Most patients (65, 86 % ) had the serum and ascites albumin levels measured < 12 h apart. Low SAAG and cirrhosis We evaluated the 76 patients with cirrhosis and at least one paracentesis yielding a SAAG level of < 1.1 g / dl. The mean SAAG level of these patients was 0.81 (s.d. = 0.21). The most common cause of cirrhosis was alcohol ( n = 35), followed by both chronic hepatitis C virus infection and alcohol abuse as an etiology of their cirrhosis in 27 patients. Six patients had only hepatitis C virus, four had cardiac cirrhosis, two had nonalcoholic fatty liver disease, one had cryptogenic cirrhosis and one patient had chronic hepatitis B virus infection. As stated in the Methods section, we used standard criteria ( Table 1 ) f the determination of low SAAG-level etiologies. Causes of low SAAG ascites were identified in 29 (38 % ) of the 76 patients ( Table 2 ). The mean SAAG level f the patients in whom a cause was identified was 0.72 (s.d. = 0.24), which was significantly lower than the mean SAAG level f patients in whom no cause was identified, which was 0.86 (s.d. = 0.18, P = 0.005). The most common etiology was SBP, which was diagnosed in 11 (38 % ) patients. The remainder had peritoneal carcinomatosis, 8 (28 % ); nephrotic syndrome, 5 (17 % ); tuberculous peritonitis, 3 (10 % ); chylous ascites, 1 (3.5 % ); and a perfated viscous (secondary peritonitis), 1 (3.5 % ). Most patients with peritoneal carcinomatosis (7, 88 % ) had suggestive abdominal imaging studies (ultrasound, CT MRI). In all, 47 (62 % ) of the 76 patients with cirrhosis and a low SAAG had no identifiable cause of the low SAAG on initial wkup ( Figure 1 ), and 33 (70 % ) of these patients had a repeat diagnostic paracentesis. Of these, 24 (73 % ) changed to a SAAG level of 1.1 g / dl; only 9 (27 % ) patients continued to have a low SAAG. Thus, in total, 23 patients with cirrhosis had an unexplainable low SAAG (including the nine patients who had a repeat paracentesis). Seven of these patients had an appropriate complete wkup. On further analysis, 16 patients had a complete wkup, except f AFB culture of the ascites fluid. Low SAAG and no cirrhosis Sixteen patients with a low SAAG had no evidence of cirrhosis. The mean SAAG level of these patients was 0.66 (s.d. = 0.32), which was significantly lower than those observed in patients Table 2. Comparison between two groups of patients with low SAAG accding to the presence of cirrhosis Cirrhosis No cirrhosis P value Number of patients Age (mean ± s.d.) 58 ± ± Male (%) 74 (97%) 9 (75%) < SAAG level (mean ± s.d.) 0.81 ± ± Cause of low SAAG identifi ed (%) peritonitis (%) Other identifi ed secondary causes (%) SAAG, serum-ascites albumin gradient. 29 (38%) 12 (75%) (14%) 1 (6%) (24%) 11 (69%) < Cirrhosis (76) Cause identified (29, 38%) No cause identified (47, 62%) Repeat paracentesis (33, 70%) No repeat paracentesis (14, 30%) Repeat paracentesis SAAG 1.1 g/dl (24, 73%) Did not revert to SAAG 1.1 g/dl (9, 27%) Figure 1. Flow diagram f fi ndings of diagnostic wkup among patients with cirrhosis and SAAG < 1.1 g / dl. SAAG, serum-ascites albumin gradient.

4 1404 Khandwalla et al. Cause identified (12) No Cirrhosis (16) Peritoneal carcinomatosis (7) Secondary peritonitis (2) Chylous ascites (1) Nephrotic syndrome (1) peritonitis (1) No cause identified (4) Figure 2. Flow diagram f results of diagnostic wkup in patients without cirrhosis and SAAG < 1.1 g / dl. SAAG, serum-ascites albumin gradient. with cirrhosis and low SAAG ( P = 0.02) ( Table 2 ). A considerably larger proption of these patients (12, 75 % ) had an identifiable cause of the low SAAG ( Figure 2 ), with peritoneal carcinomatosis being the most common (58 % ), followed by secondary peritonitis (17 % ), chylous ascites (8 % ), nephrotic syndrome (8 % ), and SBP (8 % ). DISCUSSION Low SAAG has a po predictive value in the presence of a low pretest probability f nonptal hypertension-related ascites. In 76 patients with cirrhosis and a low SAAG level, f whom the pretest likelihood f ptal hypertension ascites is high, only 38 % of patients had an identifiable cause, of which primary bacterial peritonitis accounted f 14 %. The commonly sought-after diagnoses of peritoneal carcinomatosis, tuberculous peritonitis, nephrotic syndrome, nephrogenous, and pancreatic ascites were present in only 24 % of the patients with cirrhosis and a low SAAG. In general, patients with cirrhosis and a low SAAG had SAAG values that were slightly lower than 1.1 g / dl (mean, 0.81), and of the 47 patients with no apparent etiology of a low SAAG level, 24 (51 % ) reverted back to a high SAAG level (i.e., 1.1 g / dl) on repeat paracentesis. On the other hand, patients with a low SAAG and no cirrhosis, in whom the pretest likelihood f ptal hypertension ascites is low, had a considerably lower SAAG (mean, 0.66), and most (58 % ) of the patients had an identified cause that was malignancy related. The diagnostic yield of a low SAAG accding to the presence absence of cirrhosis has not been widely addressed. F example, the Mansour Ghanaei et al. study of patients with low SAAG in Nthern Iran found that the most frequent causes of low SAAG were tuberculous peritonitis, followed by malignancy, nephrotic syndrome, and pancreatitis (15). However, concurrent cirrhosis lack of it was not categically mentioned. Nevertheless, the few published data seem to suppt our findings. In the study by Runyon et al. (1), 901 ascites samples were collected, of which 758 (84.1 % ) samples were from patients who had cirrhosis without any other secondary cause of ascites and 42 (4.7 % ) samples were classified as mixed ascites. Tuberculous peritonitis and malignancy / peritoneal carcinomatosis were the most common causes of mixed ascites. The most common reasons f a low SAAG in 42 patients with no cirrhosis were peritoneal carcinomatosis, tuberculous peritonitis, nephrogenous / nephrotic syndrome, and pancreatic ascites (1). Part of our findings pertain to patients with cirrhosis and at least one paracentesis yielding a low SAAG in which no identifiable cause explained the SAAG level of < 1.1 g / dl. We found that approximately 51 % of patients did revert to an expected SAAG of 1.1 g / dl. This is to be expected, given the only modest depression of SAAG in these patients and thus the narrow margin of err required to reach the magical cutoff of 1.1 g / dl. However, a few patients, despite a complete wkup, continued to have an unexplained low SAAG. Several patients in this group with no identifiable etiology did not have an AFB culture. On the basis of the low yield from the rest of the patients who had an AFB culture, it is reasonable to assume that most of these patients would continue to have an unexplained low SAAG level. The actual benefit from wking up a low SAAG in patients with cirrhosis is lower than the 38 % yield we observed. Most of the additional etiologies included in this yield did not require additional wkup: bacterial peritonitis was readily detected by an elevated neutrophil count and routine culture in all cases of primary and secondary peritonitis (12 patients); peritoneal carcinomatosis was highly suggested by abdominal imaging in seven patients; pre-existing nephrotic syndrome was seen in all five patients; and known pulmonary / disseminated tuberculous peritonitis was seen in one patient. Thus, the diagnostic yield may be as low as 5 %. Given that patients with cirrhosis in whom additional etiologies were diagnosed had a mean SAAG level of 0.72, a lower cutoff f SAAG is likely to have a better positive predictive value f nonptal hypertension-related etiologies. This needs to be examined in future studies. The retrospective study design is a potential limitation of the current findings, as it is subjected to variability in the accuracy and completeness of documentation of diagnostic wkup f cirrhosis, as well as low-gradient ascites. A standardized wkup would have ensured unifmity in the interpretation of the wkup of a patient with cirrhosis and a low SAAG. There was also a variance in the collection times of the serum and ascites albumin levels among the study population. However, we are unaware of a systematic bias in which wkup would be me less intense, depending on the presence of cirrhosis and, therefe the effect of the retrospective design on the study findings is unclear. Lastly, it is possible, but not likely, that addi- The American Journal of GASTROENTEROLOGY VOLUME 104 JUNE

5 Utility of Serum Albumin Gradient Ascites 1405 tional diagnostic wkup was conducted in non-va facilities and therefe missed in this study. In summary, among patients with cirrhosis found to have a low SAAG, the overall yield of a wkup f additional causes of low SAAG level is low. The standard wkup f these patients should include analysis of the ascitic fluid f cell count, bacterial culture, albumin, and total protein. A repeat measurement should be perfmed, as reversion to a SAAG of 1.1 g / dl is commonly observed. There is, however, a subset of patients who, despite having an appropriate wkup, may have an unexplained low SAAG level; this additional targeted testing includes amylase, triglyceride, AFB, cytology, and urine proteins. A small percentage of patients with unexplained low SAAG levels do not revert to a high-gradient ascites on repeat paracentesis. The utility of further wkup beyond abdominal imaging to determine the cause of low-gradient ascites in these patients is undefined and is likely to be very low. ACKNOWLEDGMENTS Th is wk was suppted in part by the Houston VA HSR & D Center of Excellence (HFP90-020) and NIH/NIDDK Center Grant P30 DK Dr El-Serag is suppted by NIH K24DK CONFLICT OF INTEREST Guarant of the article: Hashem B. El-Serag, MD, MPH. Specific auth contributions: Hashim Khandwalla and Yemi Fasakin contributed to data collection and analysis and manuscript preparation; Hashem B. El-Serag contributed to data analysis and manuscript preparation. Financial suppt: None. Potential competing interests: None. REFERENCES 1. Runyon BA, Mont ano A A, A k r iv i d is E A et al. The serum-ascites albumin gradient is superi to the exudate-transudate concept in the differential di ag nos is of as c ite s. Ann Inte r n Me d ;1 1 7 : D itt r i ch S, Yd i L M, d e Mat ro s A A. The value of serum-ascites albumin gradient f the determination of ptal hypertension in the diagnosis of as cites. Hep ato g ast ro e nte rol o g y ;4 8 : A lb a J, T re s E, Vas que z J J. S ero -as cit ic gradient of albumin: usefulness and diagnostic limitations. An Med Int 1995 ; 12 : G oy a l A K, G oy a l SK, Pok har na D S et al. D ifferential diagnosis of ascitic fluid: evaluation and comparison of various biochemical criteria with a special reference to serum ascites albumin concentration gradient and its rel at ion to p t a l pre ssu re. Trop G ast ro e nte rol ;1 0 : D as B, Achar y a U, P urohit A. C omp ar at ive ut i l it y of s e ro as c ites albumin gradient and ascitic fluid total protein f differential diagnosis of ascites. Indi an Pe d i at r ;3 5 : L au d an no OM, Bre s c i an i P, Si lv a M. D i ag no st ic efficacy of albumin gradient in different causes of ascites. Acta Gastroenterol Latinoam 1995 ; 25 : R e c t WG, R e y nol ds T B. Sup e r i it y of s e r um-ascites albumin difference over the ascites total protein concentration in separation of transudative and exudative ascites. Am J Med 1984 ; 77 : She e r TA, Runyon BA. Sp ont ane ous b a c te r i a l p e r itonit is. D i g Dis 2005 ;2 3 : Voi g t M D, Ka lv ar i a I, Tre y C et al. D i ag no st ic value of ascites adenosine deaminase in tuberculous peritonitis. Lancet 1989 ; 1 : O r t h SR, R it z E. The nephrotic syndrome. New Engl J Med 1998 ; 338 : L ane C, Brow n M, D u ns mu i r W et al. C an sp ot urine protein/creatinine ratio replace 24 h urine protein in usual clinical nephrology? Nephrology (C arlton ) ;1 1 : Runyon BA, Ho e f s JC, Mg an T R. As c it ic fluid analysis in malignancyrelated ascites. Hepatology 1988 ; 8 : Runyon BA. Amy l as e l e vels in ascitic. J Clin Gastroenterol 1987 ; 9 : C as afont F, L op e z- Ar i as M J, C re sp o J et al. C hy l ous as cites in cirrhotic and non-cirrhotic patients. Gastroenterol Hepatol 1997 ; 20 : Mans ou r- Ghanae i F, Shaf ag h i A, B ag herzadeh AH et al. L ow g radient ascites: a seven-year course review. Wld J Gastroenterol 2005 ; 11 : Study Highlights WHAT IS CURRENT KNOWLEDGE 3 There is high sensitivity f a high serum-ascites albumin gradient (SAAG) ( 1.1 g / dl) in diagnosing ascites related to ptal hypertension. 3 Patients with mixed ascites have ptal hypertension and a variable SAAG level secondary to another underlying diagnosis. WHAT IS NEW HERE 3 Low SAAG has po predictive value f cases in which the patient has a low pretest probability f nonptal hypertension related ascites. 3 In patients with cirrhosis who have a low SAAG, the overall yield of a wkup f additional causes of low SAAG level is low. 3 Repeat measurements should be perfmed; however, a small number of patients will have an unexplained low SAAG level. 3 The usefulness of further wkup beyond abdominal imaging to determine the cause of low-gradient ascites is undefined and likely to be very low.