Ascitic fluid analysis: SAAG, cell count and SBP

Ascitic fluid analysis answers two questions at once: what the fluid is, and whether it is infected. A diagnostic paracentesis takes a few millilitres of fluid from the peritoneal cavity and measures its appearance, cell count and differential, total protein and albumin (compared with serum), plus targeted tests such as culture, amylase, triglycerides and cytology. The single most useful number derived from it is the serum-ascites albumin gradient (SAAG), which sorts the causes of ascites into those driven by portal hypertension and those that are not [Ref n=1 />^[2]. ^[1]

Cirrhosis is the commonest setting, but the panel is used for any ascites: cardiac, malignant, pancreatic, tuberculous and others. About 20 to 30% of patients with cirrhosis and ascites develop spontaneous bacterial peritonitis (SBP), which is diagnosed on the ascitic neutrophil count alone and carries a high mortality if missed [Ref n=1 />. A second dangerous diagnosis, secondary bacterial peritonitis from a perforated organ, has to be separated from it because the treatment is surgical rather than medical [Ref n=1 />[Ref n=4 />. The skills practised below are: calculating the SAAG, reading the neutrophil count correctly, recognising the patterns, and knowing which extra test each pattern demands.

This page follows the current open-access guidelines from the Korean Association for the Study of the Liver (KASL, 2026) and the British Society of Gastroenterology/British Association for the Study of the Liver (BSG/BASL, 2021), whose key cut-offs are the same ones carried in the AASLD 2021 practice guidance and the EASL 2018 clinical practice guidelines for decompensated cirrhosis [Ref n=1 />^[2][5][6].

Appearance. The first result is free, because it is what you see in the syringe. Milky or chylous fluid suggests triglyceride-rich lymphatic leak. Cloudy fluid suggests infection or malignant ascites. Dark brown fluid may point to biliary perforation. Black fluid raises suspicion of pancreatic necrosis or metastatic melanoma. Blood-stained or haemorrhagic fluid (more than 50,000 red blood cells per cubic millimetre) should make you rule out malignancy first [Ref n=1 />^[3].

Cell count and differential. The ascitic white blood cell count and differential are the backbone of the panel. The polymorphonuclear leucocyte (PMN, effectively the neutrophil) count is the diagnostic gate for SBP. The total white count matters less: in one multicentre cohort a white count of 500 cells per cubic millimetre or more was how the authors defined infected ascites [Ref n=4 />. The background leucocyte population in a normal or uncomplicated cirrhotic ascites is low, so any rise is read as infection until proved otherwise. The count is affected by how it is performed (manual microscopy versus automated flow cytometry) and by haemodilution [Ref n=2 />.

Total protein. Ascitic total protein is low in cirrhosis (often under 1.5 g/dL) because the fibrous liver lets little protein out, and high (above 2.5 g/dL) with a high SAAG in cardiac ascites, where leaky hepatic sinusoids release protein-rich lymph [Ref n=2 />. A protein below 1.5 g/dL in cirrhotic ascites identifies patients at high risk of SBP and is the usual threshold for offering primary antibiotic prophylaxis [Ref n=2 />.

Albumin and the SAAG. Serum and ascitic albumin are measured on the same day, and the serum albumin is subtracted from the ascitic albumin. The result is the serum-ascites albumin gradient (SAAG). Because albumin moves with oncotic pressure, the gradient separates portal-hypertensive ascites (high gradient, 1.1 g/dL or 11 g/L or more) from ascites produced by peritoneal disease or low-oncotic states (low gradient, under 1.1 g/dL) [Ref n=1 />^[2].

Serum albumin taken at the same time. Without a same-day serum sample the gradient cannot be calculated, and older samples mislead because albumin changes over time.

Culture. When infection is suspected, ascitic fluid is drawn aseptically and inoculated at the bedside into aerobic and anaerobic blood culture bottles, which roughly doubles the bacterial detection rate compared with simply sending a plain bottle [Ref n=1 /> [Ref n=2 />. SBP is usually monomicrobial, most often Escherichia coli, Klebsiella or streptococci [Ref n=1 />. About 40% of patients with an ascitic neutrophil count of 250 per cubic millimetre or more still have a negative culture; this is called culture-negative neutrocytic ascites (CNNA) and is treated exactly like SBP [Ref n=1 />. A positive culture with fewer than 250 neutrophils is non-neutrocytic bacterascites, which may progress to SBP in over a third of cases [Ref n=1 />.

Amylase. Ascitic amylase is the test for pancreatic ascites or pseudocyst rupture. It is typically above 1000 U/L, or more than six times the serum amylase, with mean values above 4000 U/L in published series [Ref n=2 />. It should be requested when there is a history of pancreatitis or pseudocyst, or when the SAAG is low with a high protein.

Cytology. Cytological examination of ascitic fluid is particularly useful for peritoneal carcinomatosis, though the yield is highly variable, from 0 to 96.7% depending on tumour site [Ref n=1 /> [Ref n=2 />. It is at its most useful when combined with tumour markers such as carcinoembryonic antigen (CEA).

Triglycerides. A milky ascites is chylous until shown otherwise. An ascitic triglyceride above 200 mg/dL (2.26 mmol/L) supports chylous ascites, and a level below 50 mg/dL rules it out [Ref n=3 />. Chylous ascites occurs with lymphatic disruption from trauma, cirrhosis, malignancy or after surgery.

Other targeted tests. Adenosine deaminase helps separate tuberculous peritonitis from carcinomatosis, with a cut-off of under 40 U/L ruling against TB in some series [Ref n=2 />. Bilirubin supports biliary perforation. Creatinine and urea rise when ascites is really leaking urine. Ascitic BNP supports cardiac ascites. Lactate dehydrogenase and glucose are the markers for secondary bacterial peritonitis [Ref n=1 /> [Ref n=2 />.

Reference intervals are not universal. They depend on the method, the population used to set them, age, sex, method platform and the local laboratory, so the laboratory's own interval always takes priority and every result should be read against it [Ref n=1 /> [Ref n=2 /> [Ref n=5 />. The values below are illustrative teaching cut-offs from current open-access guidelines, not your laboratory's ranges.

A plain note: ranges differ by laboratory, age, sex and method, so the laboratory's own range always wins, and where a value sits relative to its cut-off matters more than the absolute number [Ref n=1 />^[2][5].

I accidentally wrote "run- criteria" typo for Runyon; fix to "Runyon criteria". Let's ensure no typo. Also "under 1.5 g/dL" etc.

The SAAG is the organising axis. High-gradient ascites (SAAG 11 g/L or more) is produced by increased hydrostatic pressure in the hepatic sinusoids, usually portal hypertension. Low-gradient ascites (SAAG under 11 g/L) arises when the peritoneum itself is diseased or albumin is low.

Interpretation of the other numbers in cause-finding:

• A cloudy or turbid ascites with a high cell count brings infection or malignancy to the front of the differential, not just cirrhosis [Ref n=1 />.
• A high protein above 2.5 g/dL combined with a high SAAG is the classic signature of cardiac ascites rather than cirrhosis [Ref n=2 />.
• A low SAAG plus a very high amylase points to pancreatic ascites, including a disrupted pancreatic duct or pseudocyst leak [Ref n=2 />.
• A low SAAG with multinucleated malignant cells or raised CEA points to peritoneal carcinomatosis [Ref n=1 />^[2].
• Tuberculous peritonitis usually gives a low SAAG with a lymphocyte-predominant count and raised adenosine deaminase [Ref n=2 />.

"Ascites causes by gradient" in short form: high gradient means the liver's sinusoids are under pressure; low gradient means the peritoneum or the albumin is the problem.

Named combinations are what make the panel useful. These are the patterns to learn:

• Uncomplicated cirrhotic ascites: high SAAG (11 g/L or more), low protein (under 1.5 g/dL), low PMN count (under 250 per mm3), negative culture. SAAG stays high and protein stays low between taps [Ref n=1 />^[2].
• SBP: high SAAG cirrhotic ascites with PMN count 250 per mm3 or more, usually in a cloudy sample, with or without a positive culture, and usually monomicrobial [Ref n=1 />^[4].
• Culture-negative neutrocytic ascites (CNNA): PMN count 250 per mm3 or more with a negative culture. This is SBP for treatment purposes, because 40% of culture-positive cut-offs are missed and the mortality is similar [Ref n=1 />.
• Non-neutrocytic bacterascites: positive culture with under 250 PMN per mm3. This may be transient colonisation, but 38% progress to SBP within days, so it warrants repeat tapping and treatment if symptoms or the count rises [Ref n=1 />.
• Secondary bacterial peritonitis: ascitic infection from an intra-abdominal source, usually polymicrobial, with a very high mortality (50 to 80%) and frequent need for surgery. Compared with SBP it tends to show localised abdominal signs, two or more organisms in culture, a very high neutrophil count, ascitic protein above 1.0 g/dL, ascitic LDH above the upper limit of the serum LDH, ascitic glucose 50 mg/dL or less, or failure of the PMN count to fall after 48 hours of appropriate antibiotics [Ref n=1 />^[4]. Older Runyon criteria (two out of three of glucose under 50 mg/dL, protein above 10 g/L and LDH above the upper limit of normal serum) are one way of picking it out, although newer criteria include also carcinoembryonic antigen above 5 ng/mL or alkaline phosphatase above 240 U/L for bowel perforation [Ref n=1 />^[4].
• Pancreatic ascites: low or variable SAAG depending on the series, but the defining number is an ascitic amylase above 1000 U/L or more than six times the serum, with mean values above 4000 U/L [Ref n=2 />.
• Chylous ascites: milky appearance, triglycerides above 200 mg/dL, and a lymphocyte-predominant cell count [Ref n=3 />.
• Haemorrhagic ascites: blood-stained with more than 50,000 red blood cells per mm3, or a meat-wash appearance; malignancy must be ruled out first [Ref n=3 />.
• Malignant ascites (peritoneal carcinomatosis): low SAAG, high protein, often cloudy, with positive cytology or raised tumour markers, and it does not respond as transudates do [Ref n=1 />^[2].
• Cardiac ascites: high SAAG with total protein above 2.5 g/dL, often with an elevated NT-proBNP measured in serum or ascites [Ref n=2 />.

The SAAG itself should not be re-measured once it has been established, as long as the clinical picture does not change: in a treated cirrhotic patient only the cell count and differential are needed on repeat therapeutic taps [Ref n=1 />.

Every sheet below is an illustrative sample invented for teaching. None of it is a real patient or a real laboratory report, and the sample values are not claims.

Sample sheet 1 (uncomplicated cirrhotic ascites). Illustrative sample only.

Step-by-step interpretation:

1. The fluid is clear and straw-coloured with a PMN count of 45 per mm3, well under the 250 per mm3 cut-off for SBP, and the total white count is under 500 per mm3 [Ref n=1 />^[2].
2. SAAG is 17 g/L (serum 33 minus ascitic 16, in g/L units), which is above the 11 g/L high-gradient threshold, and the protein of 1.1 g/dL sits below 1.5 g/dL [Ref n=1 />^[2].
3. The pattern is therefore uncomplicated portal-hypertensive ascites from cirrhosis, and the culture, amylase, triglycerides, cytology and red-cell count add nothing useful this time [Ref n=1 />^[2].
4. On future taps in a stable patient, only the cell count and differential need repeating; SAAG and protein need not be re-measured [Ref n=1 />.
5. With a protein under 1.5 g/dL this patient should be considered for primary prophylaxis against SBP, and a further fall in counts should be read as deterioration [Ref n=2 />. Conclusion: uncomplicated high-gradient cirrhotic ascites; start prophylactic antibiotics according to local policy and arrange outpatient therapeutic paracentesis for symptoms.

Sample sheet 2 (SBP in a cirrhotic patient, 62 years, febrile). Illustrative sample only.

Step-by-step interpretation:

1. The cloudy appearance and a PMN count of 640 per mm3 are well above the 250 per mm3 diagnostic threshold for SBP [Ref n=1 />^[2].
2. The total white count of 1980 per mm3 is above the 500 per mm3 level used to define infected ascites in multicentre work [Ref n=4 />.
3. Gram stain is negative, as it often is, but bedside inoculation of blood culture bottles grew E. coli, confirming a monomicrobial ascitic infection [Ref n=1 />^[2].
4. The high SAAG (14 g/L) and low protein (1.2 g/dL) show that this is SBP on top of cirrhotic ascites, not secondary peritonitis from a perforated viscus [Ref n=1 />^[2].
5. SAAG does not need re-measuring; a repeat tap in 48 hours is used to check response, with the PMN count as the marker [Ref n=1 />^[2]. Conclusion: spontaneous bacterial peritonitis; start empirical antibiotics at once and add albumin per the guidelines, with a repeat diagnostic tap at 48 hours if there is no clinical response.

Sample sheet 3 (pancreatic ascites). Illustrative sample only.

Step-by-step interpretation:

1. The protein of 3.1 g/dL and amylase of 5200 U/L dominate this report: amylase is far above the 1000 U/L or 6-times-serum threshold for pancreatic ascites [Ref n=2 />.
2. The PMN count of 80 per mm3 is well below the SBP threshold, and the culture is sterile after ten days, so this is not SBP and infection is not the problem [Ref n=1 />^[2].
3. A high SAAG (14 g/L) with a high protein is unusual for pancreatic disease, which is usually a low-gradient ascites; this discrepancy is a cue to look for two simultaneous causes, such as cirrhosis plus a pancreatic duct disruption [Ref n=2 />.
4. Negative cytology and normal triglycerides do not change the diagnosis, and AFP/CEA are not supported by the clinical picture [Ref n=1 />^[3].
5. The next step is imaging of the pancreatic duct (cross-sectional or MRCP) and, where available, endoscopic or surgical drainage of the leak or pseudocyst [Ref n=2 />. Conclusion: pancreatic ascites, probably on top of underlying chronic liver disease; arrange pancreatic imaging and surgical or endoscopic review, and repeat the tap only if the cell count rises.

• Sending an un-inoculated bottle. Culture in a plain container misses many cases; bedside inoculation of aerobic and anaerobic blood culture bottles is the expected technique [Ref n=1 /> [Ref n=2 />.
• Misreading the PMN count after a bloody tap. When ascites is heavily blood-stained, subtract 1 PMN per 250 red blood cells before applying the 250 per mm3 cut-off; otherwise a haemorrhagic tap looks like SBP [Ref n=1 />.
• Forgetting the serum albumin. Without a same-day serum sample the SAAG cannot be calculated. Using an albumin from days ago is one of the commonest causes of misclassification.
• Treating CNNA differently. A PMN count above the cut-off with a negative culture still counts as SBP, because culture negative does not mean uninfected and the mortality is similar [Ref n=1 />.
• Confusing SBP with secondary peritonitis. A low glucose (50 mg/dL or less), protein above 1.0 g/dL, LDH above the serum upper limit, polymicrobial growth, markedly raised neutrophils, or failure of the count to fall after 48 hours, point to a surgical abdomen rather than SBP [Ref n=1 />^[4].
• Reading TSH in the wrong decimal place: by analogy, ignoring units kills this panel too. SAAG is reported in g/L or g/dL and the threshold is 1.1 g/dL, i.e. 11 g/L; a decimal error flips the cause. Write the calculation (serum minus ascitic albumin) on every report.
• Overtreating with albumin. Albumin is given with large-volume paracentesis, not routinely, and not because total protein is low.
• Over-relying on amylase. A raised amylase can be salivary or from enteric rupture as well as pancreatic; always ask for lipase and confirm with imaging before assigning the label.
• Calling any cloudy fluid infected. Turbid malignant or haemorrhagic ascites can be cloudy without infection, and a PMN count below 250 argues strongly against SBP.
• Re-measuring SAAG every tap. It does not change in a stable patient and re-measuring wastes effort; re-check only if the clinical picture changes [Ref n=1 />.
• Underestimating pancreatic ascites. A normal amylase does not exclude a pancreatic cyst, and a very high amylase with a normal cell count already says pancreatic.

• A diagnostic paracentesis is mandatory for new or worsening ascites, for suspected infection, and for any patient with cirrhosis and a deteriorating admission [Ref n=1 />^[2].
• The SAAG (serum minus ascitic albumin, in g/L or g/dL) is the organising number: 11 g/L (1.1 g/dL) or more means portal hypertension, under 11 g/L means a peritoneal or hypoalbuminaemic cause [Ref n=1 />^[2].
• The ascitic PMN count is the diagnostic gate for SBP: 250 per mm3 or more, with haemodilution correction of 1 PMN per 250 red cells if the tap is bloody [Ref n=1 />^[2].
• High-gradient ascites includes cirrhosis, cardiac failure, Budd-Chiari syndrome, portal vein thrombosis, sinusoidal obstruction syndrome and massive liver metastases [Ref n=1 />.
• Low-gradient ascites includes peritoneal carcinomatosis, tuberculosis, pancreatic ascites, bowel perforation, nephrotic syndrome and myxoedema [Ref n=1 />^[2].
• Total protein below 1.5 g/dL in cirrhotic ascites marks a high risk of SBP and selects patients for primary antibiotic prophylaxis; above 2.5 g/dL with a high SAAG signposts cardiac ascites [Ref n=2 />.
• SBP is usually monomicrobial (E. coli, Klebsiella, streptococci) and is treated empirically; culture-negative neutrocytic ascites is treated the same way [Ref n=1 />.
• About 40% of SBP cases have a negative culture despite the right cut-off, so the neutrophil count governs treatment, not the culture report [Ref n=1 />.
• Secondary peritonitis carries a 50 to 80% mortality and a surgical abdomen, and is picked out by Runyon-type or extended criteria rather than by the PMN cut-off alone [Ref n=1 />^[4].
• Amylase above 1000 U/L (or more than 6 times serum) diagnoses pancreatic ascites, and triglycerides above 200 mg/dL with a level under 50 mg/dL ruling it out frame chylous ascites [Ref n=2 />^[3].
• Milky, cloudy, dark-brown or black fluid each point to a specific mechanism and should direct which extra tests to order [Ref n=1 />^[3].
• Cytology is adjunctive for peritoneal carcinomatosis, with a yield that varies from 0 to 96.7% by tumour site [Ref n=2 />.
• Re-measure only the cell count on repeat taps in a stable patient; leave SAAG and protein alone [Ref n=1 />.
• Where guidelines differ, use the most recent open-access edition your laboratory recognises; named standards used here are KASL 2026, BSG/BASL 2021, AASLD 2021 and EASL 2018 [Ref n=1 />^[2][5][6].