3.1 Study population
Fig 1 gives the study flowchart. A total of 63 (27%) out of 230 patients were eligible for primary analysis. Of these patients with decompensated liver disease related to UDCA-refractory PBC at baseline, 31 (49%) were treated with UDCA along with FF (the FF group) and 32 (51%) continued with UDCA monotherapy (the UDCA group). Nine patients in the FF group concurrent with autoimmune hepatitis while six in the UDCA group at baseline (29% vs.18%, p = 0.338); eleven patients in the UDCA group developed decompensated liver disease for a median of 36 months (range 20–73 months) after the diagnosis of UDCA-refractory PBC while ten patients in the FF group for a median of 45 months (range 24–72 months) (34% vs.32%, p = 0.859).
Baseline characteristics of the cohort are presented in Table 1. The mean patient age was 55 ± 7 years. The majority of patients were female (84%), Child-Pugh A (75%) and AMA positive (90%). FF was administrated for a median of 12 months (range 6–40 months) after the start of UDCA. The median time of exposure to FF was 36 months (range 12–108 months). The median Mayo risk score, Model for End-Stage Liver Disease (MELD) score, APRI index and FIB-4 index for the study cohort at baseline were 5.2, 1.9, 1.6 and 4.0, respectively. No significant differences except for serum gamma-glutamyl transpeptidase (GGT) levels were identified for both groups at baseline (p = 0.030). Except for patients who experienced LT or death, three patients suffered severe decompensation events during the follow-up period (e.g., large-volume ascites, acute variceal bleeding), FF was temporarily stopped and continued when the condition stabilized.
3.2 The primary end points
3.2.1 LT-free survival
Nine patients were lost in follow-up and no information about their survival was available. In the remaining 54 patients, all-cause death, liver-related death, and LT occurred in 13, 12, and 1 patients, respectively, in the UDCA group and 4, 3, and 3 patients, respectively, in the FF group. Compared to UDCA-treated cases, FF-treated cases reported a lower rate of all-cause and liver-related mortality or need for LT by study end, despite the absence of statistical differences (25% vs.50%, chi-square test, p = 0.082, log-rank test, p = 0.130, Fig 2A; 23% vs. 48%, chi-square test, p = 0.071, log-rank test, p = 0.110, Fig 2B).
Univariate analysis of factors related to LT-free survival of FF showed that four parameters were significantly linked to LT-free survival, namely age (p = 0.048), Mayo (p = 0.041), Globe (p = 0.018) and UK-PBC (p = 0.007) (Table 2). Multivariate analysis incorporating these four indicators revealed that the only independent parameter associated with LT-free survival to FF was the Globe risk score (odds ratio [OR], 2.934; confidence interval [CI], 1.046-8.235; p = 0.041). However, four of the seven patients on FF who experienced died or underwent LT (57%) achieved endpoints with ALP falling to ≤1.00× ULN during follow-up. Achieving ALP normalization at any time point was not an independent parameter associated with LT-free survival.
3.2.2 The severity of PBC
Exposure to FF was associated with a significant decrease in the Mayo and Globe risk score compared to UDCA alone. Similar results were obtained in APRI index, FIB-4 index and Forns index (Fig 3). Higher median UK-PBC risk score was observed in the UDCA group during treatment, but no significant difference was found between both groups. During follow-up, the Mayo, Globe, UK-PBC risk score, APRI index, FIB-4 index and Forns index remained stable in the FF group (p > 0.05, all).
3.2.3 Biochemical response
Of the total cohort, three patients reached normal ALP values before the enrollment of the study. Exposure to FF was associated with a significant increase in the ALP normalization rate compared to UDCA alone in the remaining 60 patients (48% vs.16%, chi-square test, p = 0.001, log-rank test, p < 0.001, Fig 4A). Of the 42 patients with evidence of decompensation prior to the diagnosis of UDCA-refractory PBC, the ALP normalization rate was obtained in 57% of additional FF-treated cases, versus only 9% of UDCA-treated cases (chi-square test, p = 0.003, log-rank test, p < 0.001, Fig 4B). Univariate analysis of factors related to the biochemical response of FF showed that five parameters were significantly linked to biochemical response, namely ALB (p = 0.041), Tbil (p = 0.018), Mayo (p = 0.012), Globe (p = 0.007) and UK-PBC (p = 0.034) (Table 3). Multivariate analysis incorporating ALB, Tbil and Mayo at baseline revealed that the only independent parameter associated with biochemical response to FF was the Mayo risk score (OR, 0.180; CI, 0.040-0.810; p = 0.025).
Fig 5 shows the dynamic changes in ALP, Tbil, ALB, alanine aminotransferase (ALT), AST and GGT. At 60 months, the median levels of ALP decreased 65% from baseline in FF-treated cases and 19% in UDCA-treated cases (p < 0.001 and p = 0.050, respectively). The median levels of Tbil and ALB in the FF group were observed to be 36% lower and 19% higher than baseline at 60 months (p = 0.126 and 0.109, respectively), while there was a tendency for serum Tbil and ALB levels to deteriorate in cases treated with UDCA. The median levels of ALT and GGT in both groups and AST in the FF group decreased progressively during follow-up (p < 0.05, all). Compared to UDCA-treated cases, significantly lower ALP levels and higher ALB levels were observed in FF-treated cases during follow-up. No significant differences in Tbil, ALT, AST, and GGT were found between both groups. Three patients discontinued FF treatment for a period of 1-2 months, then resumed with reduced ALP levels and one patient reached normal values.
3.3 The secondary end points
Adverse events are listed in Table 4. Elevations of aminotransferase and symptoms of gastrointestinal were the most frequently adverse events. Four patients experienced self-limiting nausea, abdominal pain, cramps and myalgia, and cutaneous rash in the first six months of treatment. One participant in the FF group suffered from severe fatigue at 24 months and resolved after discontinuation of FF. Severely elevated transaminase levels (ALT or AST, 5-7 x ULN) were observed in three patients at 12-24 months of treatment and gradually decreased even after continuing FF treatment with monthly monitoring, and one had concurrent autoimmune hepatitis. One FF-treated case and five UDCA-treated cases experienced first severe progression of Tbil levels (> 100 μmol/L) after enrollment (3% vs. 15%; p = 0.196); eight FF-treated cases and one UDCA-treated case reoccurred with severe progression of Tbil levels (25% vs. 3%; p = 0.013).
Fig 6 shows the dynamic changes in serum creatinine (SCr), blood urea (BU) and eGFR. Although there was a tendency for eGFR to deteriorate in cases treated with UDCA, median BU, Scr and eGFR remained stable in both groups during follow-up. No significant differences in BU, Scr and eGFR were observed between both groups. However, one 55-year-old patient with Child-Pugh B cirrhosis and normal renal function at baseline, progressed to Child-Pugh C cirrhosis with deteriorating renal function and an eGFR of 28 ml/min/1.73 m2 at 96 months of follow-up in the FF group, and the only patient with first Tbil progression after enrollment, while two participants in the UDCA group suffered from transient renal deteriorating.