About the Fibrotic NASH Index (FNI)
The Fibrotic NASH Index estimates the probability of at-risk (fibrotic) NASH — biopsy-defined steatohepatitis with a NAFLD activity score of at least 4 and fibrosis of at least F2 — from just three routine values: AST, HbA1c and HDL cholesterol. It returns a probability between 0 and 1. At or below 0.10 fibrotic NASH is effectively ruled out (sensitivity 0.89, negative predictive value 0.93); at or above 0.33 it is ruled in (specificity 0.90); the 0.10–0.33 grey zone needs a second test. Unlike FIB-4 or the NAFLD Fibrosis Score, which target advanced fibrosis alone, FNI targets the combined activity-plus-fibrosis phenotype that current MASH drug therapy is aimed at.
Formula
FNI = eˣ / (1 + eˣ), where x = −10.33 + 2.54·ln(AST) + 3.86·ln(HbA1c) − 1.66·ln(HDL)- AST
- U/L. Natural log, positive weight 2.54.
- HbA1c
- Percent (NGSP/DCCT). Natural log, positive weight 3.86 — the dominant term.
- HDL
- mg/dL. Natural log, negative weight −1.66.
- All three logs are natural logarithms (ln), not log base 10.
- The coefficients are calibrated to AST in U/L, HbA1c in % and HDL in mg/dL. If your HDL is reported in mmol/L, multiply by 38.67 first (or use the unit toggle); if HbA1c is reported as IFCC mmol/mol, convert to % (divide by 10.929, add 2.15) before entering.
- The output is a logistic probability, so it is bounded between 0 and 1 and can never be read as a fibrosis stage.
Interpreting the result
Read FNI as a triage probability with three exits. At or below 0.10 the negative predictive value is high enough (0.93 in the derivation cohort) to stand down from further NASH-specific work-up and simply manage metabolic risk, rechecking over time. At or above 0.33 the specificity is 0.90 and the patient should move to definitive fibrosis assessment — elastography, an imaging-based score such as FAST or MAST, and specialist referral — because the probability of a treatment-relevant phenotype is now meaningful. Between the two thresholds the score is genuinely uninformative and must not be treated as negative; a second-line test is required. Because the positive predictive value at rule-in was only 0.57, a high FNI is a reason to investigate, never a diagnosis on its own.
| Score | Band | What it means | Action |
|---|---|---|---|
| ≤ 0.10 | Rule-out zone | At-risk (fibrotic) NASH unlikely — sensitivity 0.89, negative predictive value 0.93 in derivation | Manage metabolic risk factors; no immediate NASH-specific work-up. Recheck as metabolic status changes |
| 0.10–0.33 | Indeterminate | Grey zone — FNI neither excludes nor confirms fibrotic NASH | Proceed to a second-line test (transient elastography, FAST or MAST). Do not treat as normal |
| ≥ 0.33 | Rule-in zone | At-risk (fibrotic) NASH likely — specificity 0.90, positive predictive value 0.57 in derivation | Definitive fibrosis staging and specialist referral; consider trial eligibility or emerging MASH therapy |
What the Fibrotic NASH Index (FNI) needs (3 inputs)
- AST (U/L)
- Aspartate aminotransferase. It enters as a natural logarithm with a positive weight, so a rising AST increases the score. Any non-NASH cause of a raised AST — recent exercise, muscle injury, alcohol, haemolysis — will inflate it.
- HbA1c (%)
- Glycated haemoglobin as a percentage (NGSP/DCCT units). It carries the largest weight of the three variables, reflecting how tightly steatohepatitis activity tracks with dysglycaemia. Enter a percentage, not IFCC mmol/mol.
- HDL cholesterol (mg/dL)
- High-density lipoprotein cholesterol. It enters with a negative weight, so a low HDL — the typical metabolic-syndrome pattern — raises the score. HDL is altered by statins, fibrates, heavy alcohol use and pregnancy independently of the liver.
Units. Enter AST in U/L, HbA1c as a percentage (NGSP/DCCT), and HDL cholesterol in mg/dL — the published coefficients are calibrated to those units. HDL reported in mmol/L must be multiplied by 38.67 (the calculator's unit toggle does this for you). HbA1c reported in IFCC mmol/mol must be converted to a percentage before entry (divide by 10.929 and add 2.15); this is an offset conversion, not a simple multiplication, so it cannot be handled by a unit switch and must be done first.
What it returns
- FNI probability
- A continuous value between 0 and 1 — the estimated probability of at-risk (fibrotic) NASH, not a fibrosis stage and not a NAFLD activity score.
- Risk zone
- Rule-out (≤ 0.10), indeterminate (0.10–0.33) or rule-in (≥ 0.33). The indeterminate zone is an instruction to test further, not a reassuring result.
How it is calculated
FNI is a three-variable logistic regression fitted directly against biopsy-defined fibrotic NASH rather than against a fibrosis stage. The authors screened routine laboratory and anthropometric variables in a biopsied cohort and found that AST, HbA1c and HDL cholesterol together captured most of the discriminatory information for the at-risk-NASH phenotype — AST standing in for hepatocellular injury, HbA1c for the dysglycaemia that drives steatohepatitis activity, and a low HDL for the atherogenic dyslipidaemia of metabolic syndrome. Those three are combined on the log scale and passed through the logistic function to give a probability. It is a statistical association model, not a mechanistic one, which is why anything that moves one of the three inputs for a non-hepatic reason degrades it.
Facts & figures
| Threshold | Purpose | Performance |
|---|---|---|
| ≤ 0.10 | Rule out fibrotic NASH | Sensitivity 0.89, NPV 0.93 |
| ≥ 0.33 | Rule in fibrotic NASH | Specificity 0.90, PPV 0.57 |
| 0.10–0.33 | Indeterminate | Second-line testing required |
The two thresholds are deliberately asymmetric: a low cut-off tuned for exclusion and a higher one for confirmation, with a grey zone between them.
| Score | Endpoint | Stage vs activity |
|---|---|---|
| FNI | At-risk NASH: NAS ≥ 4 and fibrosis ≥ F2 | Activity AND fibrosis |
| FIB-4 | Advanced fibrosis (F3–F4) | Fibrosis only |
| NAFLD Fibrosis Score | Advanced fibrosis (F3–F4) | Fibrosis only |
This is the distinction that matters for treatment: drug therapy is aimed at the activity-plus-fibrosis phenotype, not at fibrosis stage alone.
Evidence
Derivation — bariatric surgery cohort
2023 · n = 264264 patients undergoing bariatric surgery with intraoperative liver biopsy and no other cause of liver disease. AST, HbA1c and HDL were selected by logistic regression against biopsy-defined fibrotic NASH (steatohepatitis with NAFLD activity score ≥ 4 and fibrosis ≥ F2).
AUROC 0.78 (95% CI 0.71–0.85). At ≤ 0.10, sensitivity 0.89 and negative predictive value 0.93; at ≥ 0.33, specificity 0.90 and positive predictive value 0.57.
External validation in dysglycaemia
2022Applied by Pina and colleagues to cohorts stratified by glycaemic severity and type 2 diabetes duration, testing whether the HbA1c-weighted score holds up across the diabetes spectrum, and by subsequent independent metabolic-clinic cohorts.
Discrimination for at-risk NASH broadly reproduced the derivation figures, with the expected sensitivity of the score to glycaemic control noted as a caveat.
How it compares
Fibrotic NASH Index (FNI) vs FIB-4
They answer different questions — FIB-4 estimates advanced fibrosis, FNI estimates at-risk NASH — so use FIB-4 to triage fibrosis risk and FNI when the question is specifically whether treatable steatohepatitis activity is present alongside it.
FIB-4 (age, AST, ALT, platelets) is the guideline first-line rule-out for advanced fibrosis and is what most primary-care and diabetes pathways specify. FNI predicts a composite of activity and fibrosis rather than a fibrosis stage, so it can flag a patient with active NASH and early fibrosis whom FIB-4 reads as low-risk, and equally it says nothing about a burnt-out cirrhosis with low transaminases that FIB-4 would catch. They are complementary, not interchangeable.
Fibrotic NASH Index (FNI) vs NAFLD Fibrosis Score
The NAFLD Fibrosis Score targets advanced fibrosis; FNI targets at-risk NASH — pick the score whose endpoint matches your decision, and note both are blood-only and can be run over existing labs.
Both are logistic scores from routine data. The NAFLD Fibrosis Score (age, BMI, glucose/diabetes, AST, ALT, platelets, albumin) was fitted against advanced fibrosis; FNI was fitted against the activity-plus-fibrosis phenotype using metabolic markers (HbA1c, HDL). Where fibrosis staging is the goal the NAFLD Fibrosis Score or FIB-4 is the natural choice; where identifying the treatment-relevant NASH phenotype is the goal, FNI is purpose-built for it.
Fibrotic NASH Index (FNI) vs FAST score
Same endpoint, different inputs — both estimate at-risk NASH, but FNI needs only bloods while FAST needs a FibroScan (liver stiffness plus CAP plus AST), so FNI screens and FAST confirms where elastography is available.
FAST combines vibration-controlled transient elastography (liver stiffness and controlled attenuation parameter) with AST to predict the same NAS ≥ 4, F ≥ 2 target. It is more accurate than a blood-only score but requires the machine, an operator and a suitable patient. A sensible pathway runs FNI first across a metabolic population to exclude the clear negatives, then FAST or MAST on those who remain.
Pearls & pitfalls
- FNI does not measure fibrosis. It estimates the probability of a combined activity-and-fibrosis phenotype, so it answers a different question from FIB-4 or elastography and the two can legitimately disagree.
- HbA1c carries the heaviest weight. Anything that falsifies HbA1c — haemoglobinopathies, recent transfusion, haemolysis, iron deficiency, chronic kidney disease — will distort the score, sometimes markedly.
- A low HDL raises the score. Because HDL falls with statins, heavy alcohol use and poorly controlled metabolism, a very low HDL can push FNI up for reasons only partly related to the liver.
- The rule-in PPV is only 0.57. A score above 0.33 roughly doubles the pre-test odds but is not a diagnosis; it is a trigger for definitive staging.
- It was derived in a bariatric cohort with a high prevalence of NASH and extreme obesity. In a leaner, lower-prevalence clinic the positive predictive value will be lower still.
- The grey zone between 0.10 and 0.33 is common and carries no information — stopping there and calling it negative is the characteristic misuse.
- Use percentage HbA1c and mg/dL HDL. Entering IFCC mmol/mol for HbA1c, or mmol/L for HDL, without converting will give a badly wrong probability because the terms are logarithmic.
Critical actions
- Confirm the units before trusting the number: HbA1c in %, HDL in mg/dL, AST in U/L.
- Act on the indeterminate zone by arranging a second-line test rather than filing the result as normal.
- At rule-in, move to definitive fibrosis staging — elastography or an imaging-based score — and specialist referral, not straight to a NASH label.
- Check that a raised AST or a low HDL has a plausible hepatic explanation before escalating; both have common non-liver causes.
- Treat the metabolic drivers at every score level — weight, glycaemic control and cardiovascular risk — because a low FNI does not make metabolic disease benign.
- Re-run the score as metabolic status changes; a single value in someone with fluctuating diabetes control is weak evidence over years.
Why this score exists
FNI was designed around a deliberate choice of endpoint. Its authors argued that the clinically actionable target in fatty liver disease is not fibrosis stage in isolation but 'at-risk NASH' — the combination of histological activity and fibrosis that identifies patients likely to progress and eligible for intervention — and they fitted the score against exactly that composite. Choosing AST, HbA1c and HDL was equally deliberate: three values present in almost every metabolic patient's record, so the index could be deployed at population scale over existing data rather than requiring a new test. The trade-off they accepted is that a score anchored on HbA1c inherits everything that perturbs HbA1c, and that a three-variable model derived in an extreme-obesity surgical cohort carries only a modest positive predictive value into leaner clinic populations.
About the creator
First author, 2023 derivation study
Derived the index from three routine values — AST, HbA1c and HDL cholesterol — for detecting fibrotic steatohepatitis.
Senior author
Led the genetic and metabolic liver disease programme in which the index was developed.
Limitations
- It estimates a probability of at-risk NASH, not fibrosis stage or activity grade, and cannot substitute for either.
- Derived in a single bariatric-surgery cohort with high NASH prevalence and extreme obesity, which inflates the apparent positive predictive value relative to a general clinic.
- HbA1c is the dominant term, so any condition that falsifies HbA1c falsifies the score.
- HDL and AST both have common non-hepatic determinants that the model cannot distinguish from liver-driven change.
- The rule-in positive predictive value is modest (0.57), so it identifies whom to investigate rather than establishing a diagnosis.
- It has not been validated outside metabolic (MASLD/NAFLD) liver disease and should not be used in viral, alcohol-related or cholestatic disease.
- A substantial fraction of patients fall in the 0.10–0.33 grey zone, where the score provides no information.
If you are the patient
The Fibrotic NASH Index is a number worked out from three ordinary blood results — a liver enzyme (AST), your average blood-sugar marker (HbA1c) and your 'good' cholesterol (HDL). It estimates the chance that fatty liver disease has tipped over into the more active, scar-forming form that is worth treating, rather than the common quiet kind that usually just needs healthy-lifestyle changes. A low result (0.10 or under) means that active, scarring form is unlikely, and the focus stays on managing weight, blood sugar and cholesterol. A higher result (0.33 or above) means it is more likely, and you would usually be sent for a liver scan such as a FibroScan and seen by a specialist. A result in between means the blood test could not tell, so another test is needed — this is not bad news, just a sign that more information is required. Because your blood-sugar marker matters a lot in this calculation, things that affect it, such as some blood conditions, can change the number, and your team takes that into account.
Frequently asked questions
What is the Fibrotic NASH Index?#
The Fibrotic NASH Index (FNI) is a blood-based score that estimates the probability of at-risk (fibrotic) NASH — biopsy-defined steatohepatitis with a NAFLD activity score of at least 4 and fibrosis of at least stage F2 — from AST, HbA1c and HDL cholesterol. It is used to screen people with metabolic risk factors for the NASH phenotype that treatment is aimed at.
What is a normal or reassuring FNI?#
An FNI at or below 0.10 is the rule-out result: at-risk NASH is unlikely, with a negative predictive value of 0.93 in the derivation cohort. There is no single 'normal' value — the score is a probability, and the reassuring band is defined by that 0.10 threshold rather than by a reference range.
How is the Fibrotic NASH Index different from FIB-4?#
FIB-4 estimates advanced fibrosis (stage F3–F4) alone, whereas FNI estimates at-risk NASH, which combines disease activity (NAS ≥ 4) with fibrosis (≥ F2). They answer different questions, so a patient can be low-risk on one and not the other. FIB-4 is the first-line fibrosis rule-out; FNI is aimed at identifying the treatment-relevant NASH phenotype.
What does an FNI above 0.33 mean?#
It falls in the rule-in zone, where specificity is 0.90 — at-risk NASH is likely and definitive fibrosis staging plus specialist referral are warranted. It is not a diagnosis: the positive predictive value at this threshold was only 0.57 in the derivation cohort, so it identifies whom to investigate rather than confirming the disease.
Which units does the FNI calculator need?#
AST in U/L, HbA1c as a percentage (NGSP/DCCT), and HDL cholesterol in mg/dL. Because the three inputs enter as natural logarithms, using the wrong units — IFCC mmol/mol for HbA1c or mmol/L for HDL — produces a badly wrong probability. Convert first, or use the calculator's HDL unit toggle.
Can FNI replace a liver biopsy or a FibroScan?#
No. FNI is a screening tool that excludes at-risk NASH cheaply in low-probability patients and flags higher-probability ones for further testing. Those above the rule-out threshold still need definitive assessment — elastography, an imaging-based score such as FAST or MAST, or occasionally biopsy — to stage fibrosis and confirm the phenotype.
Is FNI validated in people without diabetes?#
It was derived in a bariatric-surgery cohort and has been validated across a range of glycaemic severities, including non-diabetic metabolic patients. Because HbA1c carries the largest weight, performance is most robust where glycaemia is the relevant driver; in leaner, lower-prevalence populations the positive predictive value falls, which is the main caveat to its wider use.
References
Original / primary reference
Clinical practice guidelines
- Berzigotti A, Tsochatzis E, Boursier J, et al. EASL Clinical Practice Guidelines on non-invasive tests for evaluation of liver disease severity and prognosis – 2021 update. J Hepatol. 2021;75(3):659-689.
- EASL-EASD-EASO Clinical Practice Guidelines on the management of metabolic dysfunction-associated steatotic liver disease (MASLD). J Hepatol. 2024;81(3):492-542.
- Rinella ME, Neuschwander-Tetri BA, Siddiqui MS, et al. AASLD Practice Guidance on the clinical assessment and management of nonalcoholic fatty liver disease. Hepatology. 2023;77(5):1797-1835.
- Kanwal F, Shubrook JH, Adams LA, et al. Clinical Care Pathway for the Risk Stratification and Management of Patients With Nonalcoholic Fatty Liver Disease. Gastroenterology. 2021;161(5):1657-1669.
Further reading
- Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology. 2005;41(6):1313-1321.
- Sterling RK, Lissen E, Clumeck N, et al. Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology. 2006;43(6):1317-1325.