790
Views
0
CrossRef citations to date
0
Altmetric
Research Articles

Reference change values of FIB-4

ORCID Icon, , &
Pages 394-396 | Received 24 May 2023, Accepted 24 Jul 2023, Published online: 28 Jul 2023

Abstract

When comparing two analytical results for the same analyte, the clinicians may benefit from knowing the reference change values (RCVs) of the analyte. For Fibrosis-4 Index (FIB-4), a noninvasive test used for assessing the risk of liver fibrosis, no RCVs have been published for non-cirrhotic individuals. Therefore, we estimated RCVs for adults, using retrospectively collected data from outpatients with AST, ALT, and thrombocytes within the respective reference intervals. FIB-4 was calculated as (age × AST)/(thrombocytes × ALT0.5). From two FIB-4 values in each patient we calculated the RCVs parametrically and non-parametrically. For both methods, we estimated the limits of the central 90% of the distribution of the ratio between the second and the first measurement. We obtained data on 599 outpatients with two blood tests taken 3 - 972 (median 258) days apart. The RCVs were 0.72 − 1.40 and 0.72 − 1.43, respectively, using the parametric and non-parametric methods. The 5 and 95 percentiles were not statistically significantly associated with sex, age, level of analyte, or the time between the measurements. The within-subject biological variation of FIB-4 was estimated to be 13.9%. Conclusion: In 90% of the patients the ratio between the second and the first FIB-4 result was approximately 0.7 − 1.4.

Introduction

Fibrosis-4 Index (FIB-4) is used to assess the prevalent risk of advanced fibrosis in patients with non-alcoholic fatty liver disease (NAFLD) [Citation1]. Besides, changes in FIB-4-values are associated with future severe liver disease in the general population [Citation2]. FIB-4 is calculated using age and three blood test results: FIB-4 = (age (years) × AST (U/L))/(thrombocytes (×109/L) × ALT0.5(U/L)). If the clinician needs to compare two analytical results for FIB-4 it is useful to know the reference change values (RCVs). Such data are sparse. We did not find any published values on the normal within-subject biological coefficient of variation (CVi) of FIB-4, which, along with the normal analytical coefficient of variation (CVa) could be used to calculate RCVs. However, RCVs in a population of patients with stable compensated liver cirrhosis have been published [Citation3]. Here we present RCVs of FIB-4 estimated from data in an adult Norwegian outpatient population.

Materials and methods

Population

From the laboratory information system we collected non-identifiable data from adult outpatients in the period April 2020 - January 2023. We obtained data on age, AST, ALT and thrombocytes measured in the same sample for different samples taken at least 1 day apart from the same patient. Age was calculated as the mean of age at the two sampling days. Patients where these blood tests were measured for more than 2 times during the collecting period were excluded because we thought multiple testing could indicate disease. For the same reason, we also excluded patients with any value of AST, ALT or thrombocytes exceeding their respective reference limits. The reference limits for women were 15-35 U/L of AST and 10-45 U/L of ALT, and for men they were 15-45 U/L for AST and 10-70 U/L for ALT. The reference limits of thrombocytes were 164-370 × 109/L for both women and men.

In a sensitivity analysis, we included data from patients where the relevant blood tests were measured for more than 2 times during the collecting period, using the first two FIB-4 results. In a second sensitivity analysis, we excluded data from patients with less than 30 days or more than 730 days (2 years) between the two samplings.

Analytical methods

AST and ALT were measured on Siemens Advia Chemistry XPT analyzers with reagents from the manufacturer (Siemens Healthineers AG, Erlangen, Germany). Thrombocytes were measured on Sysmex XN haematology analyzers (Sysmex, Kobe, Japan). Ordinary internal and external quality control routines were followed. The normal CVa (within-laboratory, i.e. reproducibility) of AST, ALT, and thrombocytes were 1.6%, 2.4%, and 3.1%, respectively.

Statistical methods

To estimate the total CV (CVt), i.e. (CVi2 + CVa2)0.5, we first calculated, for each patient, the difference between the second and the first measurements of FIB-4 in percent of the mean value of the two values (DPM). Then we estimated the standard deviation of the distribution of the DPMs (sDPM), after first excluding observations that were not compatible with a Gaussian distribution. In this judgement, we plotted quantile points of the DPM distribution against the corresponding quantile points of a Gaussian distribution, in addition to the Shapiro-Francia test. Finally, we divided sDPM by 20.5. This method, which is robust against systematic differences between the first and second measurement, gives the CV of each measurement, i.e. CVt [Citation4]. To estimate CVa for FIB-4, we simulated 1 million FIB-4-values, each value being calculated from the same fictive, basal figures of age, AST, ALT, and thrombocytes, but where AST, ALT and thrombocytes were randomly drawn from Gaussian distributions with the corresponding CVa from our laboratory. CVi was calculated as (CVt2 - CVa2)0.5. Parametrically, RCVs in percent were calculated as 100 × (exp(± z × 20.5 × (lnCVi2 + lnCVa2)0.5) − 1) % [Citation5], where z = 1.65 [Citation6], and lnCV = (ln(1 + CV2))0.5 [Citation7]. RCVs as ratios between the second and first measurement were calculated as (RCVs in percent + 100)/100. We also estimated the 5 and 95 percentiles in the distribution of the ratio between the second and first FIB-4 measurement, which should correspond to the parametrically estimated RCVs. We used quantile regression to study whether the 5 and 95 percentiles were associated with sex, age, the mean value of FIB-4, and the number of days between the second and first measurement. The Stata software, version 16 (StataCorp, College Station, TX 77845, USA) was used for all statistical analyses, except for simulations, where we used the software Statistics101, version 5.7 (http://www.statistics101.net/index.htm). P-values less than 0.05 were considered statistically significant.

The project was approved by the Regional Committee for Medical and Health Research Ethics (REK midt, ref. 601056).

Results

We obtained data from 599 outpatients (64% women). Some characteristics of the study population are given in . Visually, we considered one DPM value to deviate from an otherwise approximately Gaussian distribution. After excluding that single value, the p-value of the Shapiro-Francia-test was 0.25. CVt was estimated to be 14.4%, CVa 3.70%, and CVi 13.9%. Including our CVa, the RCVs of the ratio between the second and first measurement was 0.72 and 1.40, respectively. The 5 and 95 percentile (95% confidence interval) of the distribution of the ratio between the second and first measurement were 0.72 (0.69 − 0.76) and 1.43 (1.38 − 1.48), respectively. In the quantile regression models of the 5 and 95 percentiles, each model including sex, age, the mean value of FIB-4, and the number of days between the second and first measurement, none of the variables reached statistical significance.

Table 1. Characteristics of 599 outpatients, 64 % of whom were women.

Including data from patients where the relevant blood tests were measured more than two times during the collecting period and using the first two results, increased the number of patients to 940. The 5 and 95 percentile (95% confidence interval) of the distribution of the ratio between the second and first measurement were 0.73 (0.71 − 0.76) and 1.44 (1.40 − 1.49), respectively.

Excluding data from patients with less than 30 days or more than 730 days between the two sampling days reduced the number of patients to 528. The 5 and 95 percentile (95% confidence interval) of the distribution of the ratio between the second and first measurement were 0.72 (0.67 − 0.75) and 1.42 (1.38 − 1.48), respectively.

Discussion

We estimated RCVs of FIB-4 in two ways. The parametric method (0.72 − 1.40) gave slightly lower values and a narrower interval than the non-parametric method (0.72 − 1.43). The 95% confidence intervals for the RCVs estimated with the non-parametric method included the estimated RCVs of the parametric method, so the limits were not likely to be statistically significantly different. Bradley et al. [Citation3], using a parametric method with z = 1.65 (as we did), found RCVs of 0.71 − 1.47 in 28 patients with stable compensated cirrhosis measured annually for 3 years. However, after 3 years there were only 11 patients left in the study [Citation3]. They also reported on a measure called ‘coefficient of variation over time’ of 17.9%, which perhaps is comparable to our CVt of 14.4%, because it was a measure of within-subject variation and included analytical variation [Citation3]. Intuitively, we would place more trust in our non-parametric estimated RCVs for clinical work, because those estimates are not based on statistical models and are derived from a large outpatient population. Admittedly, we had no clinical diagnoses in our reference individuals; however, we did exclude all patients with values of AST, ALT, and thrombocytes outside their respective reference limits, so we expect the patients to have been reasonably stable. Interestingly, Hagström et al. [Citation2], found that a change in FIB-4 of ± 0.5 was associated with no appreciable change in the risk of severe liver disease events (their Figure 2).

It is not an option to estimate CVi from CVi-data on AST, ALT and thrombocytes, analogous to our simulation to estimate CVa, because of the correlation between the variation of AST and ALT. In this population, the Spearman correlation coefficient between the change in AST and the change in ALT was 0.63 (not shown in Results). In analytical work, however, it is reasonable to assume that the instrument measures the two enzymes independently.

If RCVs depend on sex, age, level of analyte, or the time between the measurements, their clinical use may be complicated. Fortunately, the 5 and 95 percentiles showed no associations with these variables. Also, in the sensitivity analysis excluding data from patients with less than a month or more than 2 years between the two measurements, the 5 and 95 percentiles were virtually the same. Neither did the percentiles appreciably change after including data from patients where the relevant blood tests were measured for more than 2 times during the collection period, another indication that our findings should be generalizable to other populations.

In conclusion, if FIB-4 is measured twice in outpatients with normal values of AST, ALT and thrombocytes on both occasions, one would expect that in 90% of the patients the second result would be approximately 0.7 − 1.4 of the first result.

Disclosure statement

No potential conflict of interest was reported by the authors.

References

  • Mózes FE, Lee JA, Selvaraj EA, et al. Diagnostic accuracy of non-invasive tests for advanced fibrosis in patients with NAFLD: an individual patient data meta-analysis. Gut. 2022;71(5):1006–1019. doi:10.1136/gutjnl-2021-324243.
  • Hagström H, Talbäck M, Andreasson A, et al. Repeated FIB-4 measurements can help identify individuals at risk of severe liver disease. J Hepatol. 2020;73(5):1023–1029. doi:10.1016/j.jhep.2020.06.007.
  • Bradley CR, Cox EF, Palaniyappan N, et al. Variability of noninvasive MRI and biological markers in compensated cirrhosis: insights for assessing disease progression. Eur Radiol Exp. 2022;6(1):52. doi:10.1186/s41747-022-00303-y.
  • Åsberg A, Bolann BJ. Calculating the coefficient of variation from duplicate measurements: a new method. J Appl Lab Med. 2023;8(1):220–222. doi:10.1093/jalm/jfac098.
  • Carobene A, Marino I, Coşkun A, et al. The EuBIVAS project: within- and between-Subject biological variation data for serum creatinine using enzymatic and alkaline picrate methods and implications for monitoring. Clin Chem. 2017;63(9):1527–1536. doi:10.1373/clinchem.2017.275115.
  • Aarsand AK, Díaz-Garzón J, Fernandez-Calle P, et al. The EuBIVAS: within- and between-Subject biological variation data for electrolytes, lipids, urea, uric acid, total protein, total bilirubin, direct bilirubin, and glucose. Clin Chem. 2018;64(9):1380–1393. doi:10.1373/clinchem.2018.288415.
  • Røraas T, Støve B, Petersen PH, et al. Biological variation: the effect of different distributions on estimated within-Person ­variation and reference change values. Clin Chem. 2016;62(5):725–736. doi:10.1373/clinchem.2015.252296.