Liver Fibrosis Assessment: FIB-4, Elastography, and Who Actually Needs a Specialist
How liver fibrosis is staged without a biopsy: where FIB-4 fits in the referral pathway, what elastography adds, and how to read an indeterminate score.

The liver can lose the majority of its working tissue and still keep the lights on. It manufactures albumin and clotting factors, stores glycogen, detoxifies ammonia, and processes every drug that passes through the gut - and it does all of this with enough reserve that the first symptom of advanced disease is often not pain but a swollen abdomen, a bleed, or confusion. This silence is why liver disease is discovered late, and why the most consequential decision in a primary care visit is not "is something wrong" but "is there scar tissue, and how much".
Scar tissue is fibrosis, and fibrosis is what predicts outcomes: portal hypertension, varices, ascites, liver cancer, and death. Fat in the liver, by contrast, is extremely common and usually harmless on its own. Two people can both have a fatty liver on ultrasound, and one will live forty more years without a liver problem while the other is two years from cirrhosis. The number that separates them is the fibrosis stage - not the amount of fat, not the enzyme level, and not the size of the liver.
For most of the twentieth century the only way to get that number was to take a needle sample of liver tissue. A biopsy remains the reference standard, but it is invasive, expensive, painful, occasionally serious, and - the detail clinicians rarely mention at the bedside - it samples roughly one fifty-thousandth of the organ, so two needles in the same liver can disagree about the stage. The search for a substitute produced a generation of blood-based indices and a family of imaging techniques that measure stiffness. Understanding how those tools fit together is the entire content of this guide, and the arithmetic of the most widely used index is available in the FIB-4 Calculator.
Fibrosis begins as a wound-healing response. Injured hepatocytes release inflammatory signals, hepatic stellate cells transform into myofibroblasts, and they lay down collagen in the space between sinusoids and hepatocytes. When the injury stops, the collagen can be reabsorbed; when it continues for years, the architecture itself changes. Blood can no longer reach hepatocytes through normal sinusoids, so it finds other routes, pressure rises in the portal vein, and the first varices appear. That transition - from reversible deposition to fixed architectural distortion - is what staging describes.
Pathologists score it on a five-point scale: F0 is no fibrosis, F1 is portal fibrosis, F2 is periportal or bridging fibrosis, F3 is bridging fibrosis with architectural distortion, and F4 is cirrhosis. The clinically decisive line for most decisions is F2 and above, called significant fibrosis, or F3 and above, called advanced fibrosis. The distinction matters because the interventions that change outcomes - treating the viral infection, removing the offending drug, aggressive metabolic risk management, screening for varices and liver cancer - are targeted at the upper stages.
Fat and inflammation are separate axes from scar. Steatosis is the accumulation of triglyceride droplets inside hepatocytes, visible on any ultrasound as a bright liver. Steatohepatitis adds ballooning injury and lobular inflammation, which is what drives the stellate cells to produce collagen. A patient can have marked steatosis with no inflammation and no fibrosis, or modest steatosis with active steatohepatitis and rapid scarring. This is why grading the fat tells you less than staging the scar, and why an imaging report that says "moderate hepatic steatosis" answers a different question than the one that determines the plan.
The five-point scale also carries an uncomfortable amount of measurement noise. An intraobserver agreement study would show that pathologists disagree at the boundary between adjacent stages more often than the tidy F0-to-F4 ladder suggests, and the needle itself samples a fraction of the organ - roughly one fifty-thousandth, as noted above. A biopsy that reads F2 in one lobe and F3 in another is not a contradiction; it is a sample. Non-invasive tests avoid the sampling problem entirely by interrogating the whole liver, which is one reason a well-validated score can rival a single pass of a needle despite using only four numbers.
There is a second reason staging matters, and it is easy to forget in a hepatology clinic: most people with fatty liver disease do not die of liver disease. They die of cardiovascular events. The liver finding is a marker that something systemic has gone wrong, and the visit that identifies advanced fibrosis is also the visit where blood pressure, glucose, lipids, and weight stop being background noise. Tools on this site such as the BMI Calculator and the Body Fat Calculator are not substitutes for fibrosis staging, but they belong to the same conversation about metabolic risk.
Four families of tools now sit between a routine blood panel and a biopsy needle.
Blood-based indices. These combine routine labs into a single score. FIB-4 uses age, AST, ALT and platelets. The NAFLD fibrosis score uses age, BMI, diabetes status, AST/ALT ratio, platelets and albumin. APRI uses AST and platelets alone. All three are cheap, available everywhere, and share a weakness: they were built to rule out disease in populations, not to grade a single patient precisely.
Elastography. Vibration-controlled transient elastography (VCTE) sends a mechanical pulse through the liver and measures how fast the resulting shear wave travels. Stiffer liver tissue propagates the wave faster. Magnetic resonance elastography does the same job with an MRI sequence, with better accuracy and far higher cost. Both report a stiffness value in kilopascals, and both are affected by inflammation, congestion, and how much the patient has eaten.
Because elastography measures a physical property rather than a marker of it, it is easy to treat the kilopascal value as ground truth. It is not. Acute hepatitis inflates stiffness, decompensated heart failure inflates it through venous congestion, a meal inflates it through postprandial blood flow, and the measurement fails outright in a meaningful minority of patients - obesity, ascites, and narrow intercostal spaces are the classic obstacles. A well-run service therefore reports failures and uses the median of ten valid acquisitions with an interquartile range, and a clinician reads a stiffness value the way a radiologist reads any measurement: with the technical quality attached to it.
Direct serum markers. Panels such as the enhanced liver fibrosis test measure proteins involved in matrix turnover - hyaluronic acid, tissue inhibitor of metalloproteinases, procollagen peptide. They are more expensive than FIB-4 and less available, but they do not depend on the enzyme patterns that make FIB-4 misbehave during acute inflammation.
Imaging and histology. Ultrasound, CT and MRI reveal late architectural change but are insensitive to early fibrosis. Biopsy remains the reference standard for staging and for identifying coexisting disease.
| FIB-4 | Age, AST, ALT, platelets combined | Any lab with a liver panel and CBC | First-line triage |
| NAFLD fibrosis score | Age, BMI, diabetes, AST/ALT, platelets, albumin | Same | First-line alternative |
| APRI | AST and platelets | Same | Resource-limited settings |
| VCTE | Liver stiffness in kPa | Hepatology and many primary care clinics | Second-line confirmation |
| MRE | Liver stiffness by MRI | Specialist centres | Second-line, highest accuracy |
| ELF panel | Matrix turnover proteins | Reference laboratories | Second-line blood test |
| Biopsy | Histology, stage, other disease | Hospitals | Reference standard, selected cases |
The FIB-4 index was published in 2006 by Sterling and colleagues, who were looking for a way to predict fibrosis in 832 patients co-infected with HIV and hepatitis C - a group for whom biopsy was impractical at scale.[sterling-2006] The index they derived uses four variables that appear on almost every routine panel, which is precisely why it outlived the cohort that produced it. Vallet-Pichard and colleagues validated it a year later in 847 biopsies from patients with hepatitis C alone, reporting an AUROC of 0.85 for severe fibrosis and 0.91 for cirrhosis.[vallet-pichard-2007]
The American Gastroenterological Association pathway places FIB-4 first for a structural reason: it costs nothing beyond tests already ordered, so it can be applied to every patient with a fatty liver without a prior decision about who deserves testing.[kanwal-2021-aga] The AASLD practice guidance describes the same sequence - a non-invasive score in everyone, a second-line test in those who are not clearly low or clearly high, and referral for those who remain elevated.[rinella-2023-aasld]
The individual patient data meta-analysis by Mózes and colleagues is the best single source for how well this works. Pooling 37 studies, FIB-4 reached an AUROC of 0.76 for advanced fibrosis, the NAFLD fibrosis score 0.73, and elastography 0.85. Used in sequence - FIB-4 below 1.3 to rule out, FIB-4 above 2.67 to rule in, elastography for the band in between - the pathway achieved 66% sensitivity and 86% specificity for advanced fibrosis.[mozes-2022]
| FIB-4 alone | 0.76 | Cheapest, first in line |
| NAFLD fibrosis score alone | 0.73 | Needs BMI, diabetes status, albumin |
| VCTE elastography alone | 0.85 | Best single non-invasive test |
| FIB-4 then VCTE, sequential | 0.83 | The published pathway |
| Liver biopsy | Reference standard | Invasive, sampling error |
The lesson in that table is the one that matters for resource allocation: a nearly free test followed by a moderately expensive one gets within two points of the expensive test used alone. That is why pathways and not single tests are the unit of design.
Between a third and a half of real FIB-4 results land between the cut-offs, and this is not a defect. A triage test that never produced an ambiguous middle would have to be so accurate that it would no longer be cheap.
What the indeterminate band means is precisely what it says: the index cannot separate a person with advanced fibrosis from a person without it. The correct response is a different test, not a repeat of the same one. Elastography is the usual choice; VCTE cut-offs of 8.0 and 10.0 kPa were the ones evaluated in the sequential pathway.[mozes-2022] Where VCTE is unavailable, a direct serum marker panel or referral to a centre that has one is reasonable.
Two clinical habits make the band easier to manage. First, interpret the index alongside the platelet trend: a FIB-4 that has climbed from 1.1 to 1.9 over three years is more concerning than a stable 1.6, even though both are indeterminate today. Second, read the enzyme pattern: an AST-dominant ratio in a patient who does not drink heavily suggests something other than simple steatosis.
Age appears in the numerator of the formula because fibrosis prevalence rises with age, which improves discrimination across a population. The side effect is that the same labs produce a higher score in an older person, and that the specificity of the upper cut-off erodes with age.
McPherson and colleagues quantified the erosion in 634 patients with biopsy-confirmed disease. Above age 35, FIB-4 discrimination was stable, with AUROCs between 0.77 and 0.84. Specificity at the upper cut-off, however, declined with every decade and reached roughly 35% at 65 and older - meaning about two of every three positive results in that group were false positives. An alternative upper cut-off of 2.0 restored specificity to 70% while keeping sensitivity at 77%.[mcpherson-2017]
At the other extreme the test is weak: in patients 35 and younger the AUROC fell to 0.60. A low FIB-4 in a young adult is reassuring mainly because advanced fibrosis is rare at that age.
| 35 and younger | 0.60 | Poor discrimination |
| 36 to 64 | 0.77 to 0.84 | Acceptable |
| 65 and older | 0.77 to 0.84 | Specificity about 35%; 2.0 proposed instead |
This is why a calculator that silently swaps thresholds is dangerous and one that reports which threshold it applied is not. The FIB-4 Calculator prints both the cut-offs used and the age note, so the reasoning stays visible.
Abstract thresholds become concrete when three people walk into the same clinic with the same ultrasound report.
Patient A is 42. AST 28 U/L, ALT 44 U/L, platelets 260 × 10⁹/L. The index is 42 × 28 ÷ (260 × √44) = 1176 ÷ 1724.6 = 0.68. Below 1.3, so advanced fibrosis is unlikely, and the correct action is to document the low-risk result, address the metabolic risk factors, and re-check in a few years rather than order imaging today. Note how much the ALT helped: the square root of a high ALT shrinks the denominator's growth, so an inflamed liver with an AST below ALT tends toward a lower score.
Patient B is 58. AST 52 U/L, ALT 38 U/L, platelets 168 × 10⁹/L. The index is 58 × 52 ÷ (168 × √38) = 3016 ÷ 1035.6 = 2.91. Above the 2.67 upper cut-off, so this is a high-risk result and referral is the pathway's answer. Two features of this panel deserve attention: the AST is higher than the ALT, the inverted ratio that alcohol, advanced fibrosis, and several other conditions produce, and the platelets are at the low end. Both push the numerator up and the denominator down at the same time.
Patient C is 71. AST 34 U/L, ALT 41 U/L, platelets 188 × 10⁹/L. The index is 71 × 34 ÷ (188 × √41) = 2414 ÷ 1203.8 = 2.01. Under the standard 1.3 and 2.67 cut-offs this is indeterminate, and a pathway would send this patient for elastography. Under the age-adjusted upper cut-off of 2.0 proposed for patients 65 and older, the same number crosses into the high band.[mcpherson-2017] Nothing about the patient changed between the two readings - only the threshold did, which is exactly why the threshold has to be visible in the output and not buried in a reference table.
Three panels, three different actions, and not one of them required a needle.
In 2023 a multisociety Delphi process replaced "non-alcoholic fatty liver disease" with "metabolic dysfunction-associated steatotic liver disease", or MASLD, and introduced metabolic dysfunction-associated steatohepatitis for the inflammatory form.[rinella-2023-nomenclature] The change was driven by stigma and by the recognition that the diagnosis should be defined positively - by cardiometabolic criteria - rather than by the absence of alcohol.
For anyone reading older literature or a report printed before 2024, the terms are effectively continuous: NAFLD maps onto MASLD, and NASH onto MASH, with the caveat that the new definition requires at least one cardiometabolic criterion. Staging tools did not change. FIB-4, the NAFLD fibrosis score, and elastography thresholds were validated under the old labels and remain in use under the new ones.[rinella-2023-aasld]
A fibrosis stage changes four decisions.
Referral. Advanced fibrosis or cirrhosis belongs in specialist care, where surveillance for varices and liver cancer is organised.
Risk-factor treatment. Glucose, lipids, blood pressure and weight stop being generic advice and become part of a liver-specific plan. Dietary composition matters here, and the Sugar in Food Calculator exists for the narrower question of how much added sugar a given product contributes.
Medication review. Some drugs need dose adjustment in advanced disease, and hepatotoxic combinations deserve a second look.
Vaccination and screening. Hepatitis A and B vaccination, and in some pathways liver cancer surveillance, follow from the stage rather than from the enzyme level.
Conversely, a low FIB-4 changes the plan by subtraction: no imaging today, no referral today, and a re-check interval driven by metabolic risk. That is a real clinical output, not a null result - it prevents a cascade of unnecessary scans in the majority of patients who will never develop advanced disease.
The pathway that follows from the evidence above is short enough to fit on one line of a note.
- Confirm the fatty liver finding and check that no other cause explains it - alcohol history, viral serology, medication review, and a look at the biliary tree.
- Order or retrieve AST, ALT, platelet count, albumin, and the metabolic panel. These are almost always already available.
- Compute FIB-4. Below 1.3, document low risk and set a re-check interval. Above 2.67, refer.
- In the indeterminate band, order VCTE. If VCTE is between 8.0 and 10.0 kPa, the ambiguity survives and a serum marker panel or specialist opinion is the next step.
- Whatever the stage, treat the metabolic risk. This is the intervention with the largest effect on overall mortality and the one most often skipped in a liver-focused visit.
Nothing in this sequence requires a biopsy, and nothing in it requires a specialist until the numbers say so.
After the referral. Referral is not the end of the pathway, it is a second, more expensive filter. A hepatology assessment typically repeats the history and adds the causes that primary care screening does not cover routinely - viral serology, autoimmune markers, iron studies, alpha-1 antitrypsin, and a careful alcohol and supplement history. Then it revisits the stage with a tool the first visit did not have, usually elastography where it was not already done, sometimes MRI-based liver fat quantification, and in selected cases a biopsy.
That second look changes the answer more often than patients expect. Two patterns recur: a score inflated by a transient factor that has since resolved, and a genuinely advanced stage in someone whose enzymes were never dramatically abnormal. Enzyme levels and fibrosis stage are only loosely correlated, which is the single most counterintuitive fact in this field and the reason staging tools exist at all.
Once a stage is established, the follow-up schedule follows from it: surveillance intervals for varices and for liver cancer in cirrhosis, vaccination, medication review, and the treatment of metabolic risk with the intensity that a chronic disease deserves.[rinella-2023-aasld] For most patients the referral also produces something less tangible but equally useful - a clear statement of what is being watched and why, which is what makes a long asymptomatic disease tolerable to live with.
Non-invasive tests estimate fibrosis; they do not measure it. Every threshold in this guide comes from studies with confidence intervals, and any individual result sits somewhere inside one. Discordance between a score and the clinical picture is information, not noise: it usually means the score's assumptions do not hold for that patient.
Situations where the assumptions commonly fail include acute inflammation of any cause, haemolysis or blood transfusion affecting enzyme levels, platelet disorders unrelated to the liver, congestion from heart failure, and heavy alcohol use, which raises AST and lowers platelets independently of fibrosis.[niddk-nafld] Cholestatic disease, autoimmune hepatitis, and iron overload have their own pathways and their own indices; FIB-4 was not developed for them.
This guide is educational. It explains how the tools work, what the published thresholds mean, and where the evidence is thin. It is not a diagnostic algorithm, it cannot see a patient, and it should not be used to decide anyone's care without a clinician who can review the record, examine the patient, and order the test that the score only points toward.
- ❓ Is FIB-4 better than a FibroScan?
- ✅ No. Elastography is the more accurate test, with an AUROC around 0.85 against roughly 0.76 for FIB-4. FIB-4 is used first because it costs almost nothing and resolves most patients without imaging.
- ❓ Can fibrosis reverse?
- ✅ Yes, up to a point. When the cause is removed - viral cure, alcohol cessation, weight loss - collagen can be reabsorbed and stiffness can fall. Advanced architectural change is less reversible, which is why early detection matters.
- ❓ Do I need a biopsy if my FIB-4 is high?
- ✅ Usually not immediately. A high score typically leads to elastography or a serum marker panel first, and biopsy is reserved for cases where non-invasive results conflict or another disease needs to be excluded.
- ❓ Why does my score differ from an online calculator?
- ✅ Check the platelet unit and the square root placement first. A platelet count entered as 200000 instead of 200, or a denominator that square-roots the whole product, changes the result substantially.
- ❓ Does a fatty liver always mean liver disease?
- ✅ No. Steatosis is common and often stable. The question that decides prognosis is whether fibrosis is present, which is why staging tools rather than fat quantification drive referral.
- ❓ What is MASLD?
- ✅ Metabolic dysfunction-associated steatotic liver disease. It replaced NAFLD in 2023 and is defined by steatosis plus at least one cardiometabolic criterion, rather than by the absence of alcohol.
- ❓ How much weight loss changes fibrosis?
- ✅ Published cohorts associate roughly 7 to 10 percent body weight loss with histological improvement in steatohepatitis, and greater loss with regression of fibrosis. The relationship is not linear and varies between patients.
- ❓ Is FIB-4 valid in hepatitis B?
- ✅ It is used in hepatitis B, but it was developed in hepatitis C and HIV co-infection and validated mainly in hepatitis C and fatty liver disease. Performance in hepatitis B varies by study and by viral activity.
- ❓ Can I use these numbers without a doctor?
- ✅ You can understand them, which is the point of this guide. You cannot safely act on them alone, because the interpretation depends on context the score never sees.
- ❓ What is a normal liver stiffness value?
- ✅ In VCTE, values below about 8 kPa are usually considered low risk for advanced fibrosis, 8 to 10 kPa is indeterminate, and above 10 kPa suggests advanced fibrosis. Cut-offs differ slightly by cause and guideline.
- ❓ Does alcohol show up in FIB-4?
- ✅ Indirectly and often. Alcohol raises AST more than ALT and can reduce platelets, both of which inflate the score, so a high result may reflect drinking rather than fibrosis.
- ❓ How often should fibrosis be reassessed?
- ✅ There is no universal schedule. Reassessment every one to three years is common in metabolic liver disease, with frequency driven by diabetes, obesity, and the trend of the score rather than by a fixed rule.
References
- [1]Sterling, R.K., Lissen, E., Clumeck, N., et al. (2006). "Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection." *Hepatology*, 43(6), 1317–1325.
- [2]Vallet-Pichard, A., Mallet, V., Nalpas, B., et al. (2007). "FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. Comparison with liver biopsy and FibroTest." *Hepatology*, 46(1), 32–36.
- [3]McPherson, S., Hardy, T., Dufour, J.F., et al. (2017). "Age as a confounding factor for the accurate non-invasive diagnosis of advanced NAFLD fibrosis." *American Journal of Gastroenterology*, 112(5), 740–751.
- [4]Mózes, F.E., Lee, J.A., Selvaraj, E.A., et al. (2022). "Diagnostic accuracy of non-invasive tests for advanced fibrosis in patients with NAFLD: an individual patient data meta-analysis." *Gut*, 71(5), 1006–1019.
- [5]Kanwal, F., Shubrook, J.H., Adams, L.A., et al. (2021). "Clinical care pathway for the risk stratification and management of patients with nonalcoholic fatty liver disease." *Gastroenterology*, 161(5), 1657–1669.
- [6]Rinella, M.E., Neuschwander-Tetri, B.A., Siddiqui, M.S., et al. (2023). "AASLD practice guidance on the clinical assessment and management of nonalcoholic fatty liver disease." *Hepatology*, 77(5), 1797–1835.
- [7]Rinella, M.E., Lazarus, J.V., Ratziu, V., et al. (2023). "A multisociety Delphi consensus statement on new fatty liver disease nomenclature." *Journal of Hepatology*, 79(6), 1542–1556.
- [8]National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). (2024). "NAFLD & NASH."
UnByte — Independent Software Engineering
All reference data cites its sources — Editorial policy
