Formula
RI = 1 if Creatinine > 115 μmol/L, else 0
Bilirubin Evolution = Bilirubin Day 0 − Bilirubin Day 7
R = 3.19 − (0.101 × Age)
+ (0.147 × Albumin g/L)
+ (0.0165 × Bilirubin Evolution)
− (0.206 × Renal Insufficiency)
− (0.0065 × Bilirubin Day 0)
− (0.0096 × PT seconds)
Lille Score = e^R / (1 + e^R)
Logistic regression model. Score ranges 0–1. Validated in patients with Maddrey DF ≥32 receiving prednisolone 40 mg/day.
Interpretation
| Lille Score | Response | 6-Month Mortality | Action |
|---|
| <0.16 |
Complete Responder |
<15% |
Continue prednisolone |
| 0.16–0.45 |
Partial Responder |
~35% |
Clinical judgment |
| >0.45 |
Non-Responder |
~75% |
Stop prednisolone |
| >0.56 |
Very Poor Prognosis |
>80% |
Consider liver transplant evaluation |
Clinical Application
The Lille Score is applied in the context of severe alcoholic hepatitis (Maddrey DF ≥32) in patients already receiving prednisolone 40 mg/day for the first 28 days of treatment. It is calculated on Day 7 — early enough to prevent futile steroid continuation and the associated risks of prolonged immunosuppression.
The STOPAH trial (Thursz 2015) demonstrated an overall 28-day survival benefit for prednisolone that was not sustained at 90 days or 1 year. The Lille Score identifies the subgroup who genuinely benefits — complete and partial responders — allowing the others to avoid the risk of continuing steroids without benefit.
- Glasgow Alcoholic Hepatitis Score (GAHS): Can be used alongside Maddrey DF to decide who should receive corticosteroids — GAHS ≥9 identifies highest-risk patients likely to benefit from steroids.
- Lille determines response: Once steroids are started, the Lille Score at Day 7 is the validated tool to determine whether to continue or stop.
- ABIC Score: An alternative baseline prognostic score (Age, Bilirubin, INR, Creatinine) that stratifies patients at presentation, complementary to the Lille Score's Day 7 role.
Why Stop Non-Responders
Continuing prednisolone in a Lille >0.45 non-responder offers no survival benefit while significantly increasing the risk of serious infections, including:
- Bacterial infections — spontaneous bacterial peritonitis, pneumonia, bacteraemia
- Fungal infections — invasive candidiasis, aspergillosis
- Reactivation of latent tuberculosis or hepatitis B
In non-responders (Lille >0.45), current options include:
- Stop prednisolone immediately
- Pentoxifylline: Limited evidence; the STOPAH trial found no benefit over placebo at 90 days when used as primary therapy or rescue
- N-acetylcysteine: May reduce short-term (30-day) mortality as an adjunct but evidence is limited
- Early liver transplantation: Strongly considered for highly selected first-episode non-responders without active alcohol use — evidence from Mathurin 2011 and subsequent series
Limitations
- Requires Day 7 assessment — cannot be applied at presentation
- Bilirubin may initially worsen before improving in some true responders ("early cholestatic phase") — use clinical judgment alongside the score in borderline cases
- Prothrombin time variability between laboratories; ensure consistent PT methodology
- Validated specifically in prednisolone-treated patients; not validated for pentoxifylline response assessment
References
- Louvet A, et al. The Lille model: a new tool for therapeutic strategy in patients with severe alcoholic hepatitis treated with steroids. Hepatology. 2007;45(6):1348–1354.
- Mathurin P, et al. Prednisolone with vs without pentoxifylline and survival of patients with severe alcoholic hepatitis (STOPAH): a randomised trial. Lancet. 2015;385(9958):1159–1167.