Journal of Nutraceuticals and Food Science

Keywords

Probiotics; NAFLD; REVEAL study; Epatrex; Hamaguchi score; Cumulative ordinal logistic regression

Introduction

Non-Alcoholic Fatty Liver Disease (NAFLD) is characterized by accumulation of lipids in hepatocytes at a percentage greater than 5% of the total number of cells [1]. This disease is similar to what is observed in chronic alcohol users, but occurs in patients with no significant alcohol intake (<20 g of alcohol per day). NAFLD is currently considered the most common hepatic disease in Western countries, with prevalence from 5.7%-30% in the general population, while in obese patients the prevalence is even higher and ranges from 50%-90% [2-5]. Some studies have also shown a gender-based difference in the prevalence of the disease, with greater incidence in men (2-3 times higher) among younger patients and higher prevalence in women among patients older than 60 years of age [6]. NAFLD progression may lead to Nonalcoholic Steatohepatitis (NASH) and finally to liver cirrhosis, which is associated with increased risk of tumor and other complications [7].

Until the early 2000s, liver biopsy was the most cost-effective and sensitive test in patients with NAFLD for exclusion of steatohepatitis and its possible complications. However, due to ethical concerns, needle biopsy could not be performed as a screening method to detect NAFLD in the general population. To face this problem, Hamaguchi and colleagues were looking for a reliable noninvasive method for early detection of NAFLD and in 2007 proposed a new diagnosis criterion by scoring liver ultrasonographic findings [8]. They also analyzed the sensitivity and specificity of the new score and obtained significant results, 91.7% for sensitivity and 100% for specificity, paving the way to its adoption as a noninvasive method in the detection of NAFLD. A recent study confirmed the greater accuracy of ultrasoundbased techniques with respect to biochemical indexes, finding Hamaguchi score’s sensitivity and specificity equal to 82.2% and 100.0%, respectively [9].

Different pharmacotherapeutic strategies have been assessed to treat NAFLD, but there are currently no established pharmaceutical or nutraceutical therapies and available guidelines only suggest lifestyle changes [10-12]. Clinical trials are being performed to evaluate different molecules, such as pioglitazone, metformin and Vitamin E, with some results on improved liver enzyme levels and histological features of NASH [13,14]. A review by Ni, et al. suggests that dietary antioxidant compounds, such as carotenoids which possess anti-inflammatory properties, could be used to prevent or treat NAFLD [15]. However, so far it remains unclear whether micronutrient supplementation can be used as an adjunct therapy to lifestyle changes to prevent and treat NAFLD and additional studies on new therapies are needed.

Today, clinical trials on nutraceuticals constitute a small number of studies with respect to the amount of registered studies on drugs and biologics [16]. This is due to:

• The need of a larger sample size to conduct a viable study compared to a normal drug trial.

• Higher drop-out rates than pharmaceutical trials as the study group must maintain a healthy lifestyle and record more related information.

• Need to do a pilot study to gain insight into desired end results, which results in additional costs [17].

This makes the performance of clinical studies on nutraceuticals a big challenge as well as a great value in the development of new products, especially for small-medium companies.

To overcome these challenges, data analysis is important in finding clinical evidence of the safety and efficacy of the compounds.

Computational analysis methods, which leverage computational power to extract information from complex datasets, has reached a consolidated use in life science and specifically to support the different processes of the whole product lifecycle from discovery to post-market assessment of pharmaceutical compounds [18].

Today, integrated computational solutions can help research centers, heath institutions and pharmaceutical companies accelerate and reduce costs of new products development processes [19]. Similarly, nutraceuticals producers/manufacturers can benefit from computational tools to obtain insights into a nutraceutical’s mechanism of action, investigate its safety and/or efficacy and optimize study design.

In this work, we present the statistical analyses performed to evaluate the effects of Epatrex, a novel nutraceutical for the treatment of NAFLD. The safety and efficacy of Epatrex vs. placebo have already been investigated in subjects with confirmed diagnosis of NAFLD enrolled in the REVEAL clinical trial [20]. The results of this study clearly demonstrated that the investigated nutraceutical was safe. Nevertheless, no clear differences of Epatrex over placebo with respect to the trial efficacy endpoints (laboratory exam values and markers of fatty liver, including γ-glutamyl transferase and fatty liver index) could be identified. The authors supposed that failure to demonstrate efficacy was probably due to the patients’ highly heterogenous baseline conditions and/or insufficient duration of treatment period [20]. Since expectation of epatrex efficacy is based on fairly strong assumptions emerging from the analysis of existing hepatoprotective nutraceuticals, its efficacy is worth further investigation [20]. To this purpose, we expanded the previous statistical analyses, by analyzing an additional clinical endpoint, i.e., the Hamaguchi score and by evaluating the influence of study design as well as patient specific characteristics which may represent relevant confounding factors.

The rationale for choosing the Hamaguchi score lies in its high accuracy, noninvasive nature, and practicality, making it a very useful tool for early detection and screening of NAFLD, facilitating wider adoption in clinical practice and large-scale epidemiological studies.

Our findings suggest that the Hamaguchi score is most sensitive to treatment effect, as well as patient’s covariates and study design factors and may thus be evaluated as endpoint for efficacy characterization of hepatoprotective nutraceuticals.

Materials and Methods

Treatment

Epatrex is a nutraceutical compound formulated as a threelayer tablet containing Lactobacillus casei (LC-XCAL) (109 CFU), silymarin/silybin (90 mg) and chromium picolinate (100 μg). Study participants were supplied either with Epatrex or placebo tablets on separate occasions for 2 months with a one-month washout period. Participants were asked to take two tablets per day after meals during treatment periods. The composition of the layers for both Epatrex and placebo tablets is detailed in [20].

Study design and participants

The REVEAL clinical study was a randomized, single-blind, twoperiod, two-sequence crossover trial, comparing Epatrex vs. placebo in adult subjects with confirmed diagnosis of NAFLD [20]. Subjects enrolled in this study were randomly assigned (allocation ratio 1:1) to one of the two treatment sequences (AB and BA). Subjects in sequence AB received Epatrex 1 tb/day for 60 days followed by one month of washout, then crossover period of 1 tb/day for 60 days. Subjects in sequence BA received treatment in the opposite order to sequence AB. A schematic representation of the study design is provided in Figure 1.

Figure 1: Flowchart of REVEAL study crossover phases. Enrolled subjects were randomly allocated to one of the two treatment sequences. The study lasted a total of 150 days: 60 days on Epatrex or placebo followed by 30 days of washout and continuing for another 60 days in cross-over therapy.

This study was conducted in 5 gastroenterology medical centers 1). Gastromedica SRL, Iasi, 2). Municipal hospital Dr. Gavril Curteanu oradea, 3). “Prof. Dr. Octavian Fodor” Regional Institute of Gastroenterology and Hepatology, Cluj Napoca, 4). University Emergency Hospital Bucharest, Department of Internal Medicine, Bucharest and 5. Endodigest Medical Clinic oradea) in compliance with the revised Declaration of Helsinki, Good Clinical Practice Guidelines, Certified Performance Management Professional (CPMP/ICH/135/195; ICH Topic E6) and local regulations. The protocol was approved by each ethics committee of the medical centers involved in the study. All subjects provided written informed consent before screening for eligibility.

The inclusion criteria for the trial were 18-75 years of age, confirmed diagnosis of NAFLD grade 1-4 by ultrasound, Alanine Aminotransferase (ALT) or Aspartate Aminotransferase (AST) >40 IU/L or γ-Glutamyl Transferase (GGT) >35 IU/L. The exclusion criteria for the trial included alcohol intake >20 g per day, use of nutraceuticals or antioxidant products within the last month, Body Mass Index (BMI) ≥ 40 kg/m², evidence of liver decompensation or renal insufficiency and other relevant pre-existing medical conditions (including type I diabetes, cancer, chronic immunemediated diseases).

During the study, subjects attended 4 on-site visits, where they underwent laboratory and clinical evaluation procedures described in detail in [20]. Subject demographic characteristics and information regarding alcohol consumption were collected at the first clinical visit (Visit 1-day 0). Physical measurements, laboratory exams and hepatic ultrasound measurements were longitudinally collected at the first clinical visit (Visit 1- day 0), at the end of the first treatment period (Visit 2- day 60), at the end of wash-out (Visit 3- day 90) and at the end of study (visit 4- day 150). An overview of the data types collected throughout the REVEAL study is provided in Table 1.

Table 1: Data collected throughout the REVEAL study.

Analysis dataset creation and definition

Creation of the analysis dataset was conducted using R version 4.1.2 [21]. Subjects were excluded from the analyses if they did not complete the study through Visit 4 or if the values of the required covariates were missing or deemed outliers. BMI was calculated using the standard formula, i.e., body weight in kilograms divided by the square of height in meters. Waist Circumference (WC) was categorized using the cut-points for abdominal obesity [22], into the normal (men, ≤ 102 cm; women, ≤ 88 cm) and high (men, >102 cm; women, >88 cm) WC categories. Information about alcohol consumption habits extracted from the questionnaire were alcohol consumption (regarded as a binary variable, i.e., yes/no) and age of first alcohol intake.

The main objective of the present analyses was to investigate the impact of treatment, study design and subject’s specific factors on the Hamaguchi score endpoint. This score takes values from 0 to 6 and is obtained from the following abdominal ultrasonography variables: a) liver brightness graded from 0 to 3; b) diaphragm attenuation graded from 0 to 2; and c) vessel blurring graded from 0 to 1. The total score is computed as the sum of scores a, b, c, if score a) is equal to or greater than 1. Otherwise, it is set to zero [8].

Statistical analysis

The Hamaguchi score endpoint was analyzed by cumulative ordinal logistic regression. The analyzed outcome was the Hamaguchi score change from baseline, calculated as the difference between the endpoint values at Visit 2 and Visit 1 for period 1 and the difference between the endpoint values at Visit 4 and Visit 3 for period 2. The change was considered as ordinal response variable in cumulative logit model. To formalize this, let denote the endpoint change from baseline in patient k, in sequence i (i=1 AB, i=2 BA) and period j=1,2. The model employed for th,e main analysis was defined as follows:

where

Hamaguchi score _jk is the endpoint value at Visit 1 if period j=1 and the endpont value at Visit 3 if period j=2.

X_treat, X_period, X_sequenceare indicator variables defining treatment (Epatrex or Placebo), period (period 1 or 2) and sequence (AB or BA), respectively. N is the total number of ordinal categories.

The model employed for the main analysis was then expanded to evaluate effects of covariates on the Hamaguchi score endpoint. Investigated covariates included subject’s age, gender, body size measurements (i.e., BMI and WC) and information on alcohol consumption extracted from the alcohol intake questionnaire.

Intercept parameters, and regression parameters associated to study design and patient-specific covariates were estimated in Statistical Analysis System (SAS version 9.4) by using the GENMOD procedure with the Generalized Estimating Equations (GEEs) option. Parameter estimates were reported with 95% confidence intervals. Statistical significance was set as p ≤ 0.05.

Results

Analysis dataset

A total of 74 subjects were enrolled and randomized in this study. Of these 74 patients, seven did not complete the study through Visit 4 and were thus excluded from the analysis population. The final analysis dataset contained 67 patients randomized to AB (n=33) or BA (n=34) sequences.

Summary statistics of demographic, physical measurements and ultrasound Hamaguchi score at baseline Visit 1 for the analysis population are provided in Table 2.

Table 2: Analysis population characteristics at baseline Visit 1. Values are expressed as median (min, max) or n (%).

Analysis of the Hamaguchi score endpoint

To determine whether there was an association between epatrex treatment and Hamaguchi score changes from baseline, multivariate regression analyses were performed by using the cumulative logit model for ordinal data. The analyses investigated the influence of study design factors as well as subjects’ specific covariates which may represent important confounders. Results of the cumulative ordinal logistic regression incorporating subject’s gender, age, WC category and alcohol consumption habits are reported in Table 3.

Table 3: Results from the cumulative ordinal logistic regression analysis of the Hamaguchi score endpoint in the sample of 65 subjects with known age of first alcohol intake. Baseline Hamaguchi score value at baseline visits (visit 1 for period 1, visit 3 for period 2); for categorical covariates, the category between brackets is the one associated to the regression coefficient.

Although treatment was not identified as statistically significant (p=0.7537), the interaction term for Treatment and WC (Treatment*WC) was found to be statistically significant (p=0.0045), indicating different impact of treatment in subjects of normal and high waist circumference categories. Specifically, Epatrex appeared to have a more pronounced effect on patients with high waist circumference as compared to patients with normal waist circumference values (A: High WC vs. A: Normal WC) Figure 2A. In the high WC category, Epatrex appeared to have a stronger effect compared to Placebo (A: High WC vs. B: High WC) Figure 2A, whereas a stronger effect of Placebo compared to Epatrex was observed in the Normal WC category (B: Normal WC vs. A: Normal WC) Figure 2A.

Concerning influence of study design factors, treatment sequence did not affect Hamaguchi score changes (p=0.6863), whereas treatment period was identified as strongly significant (p=0.0042). Specifically, period 1 was associated with 0.6 times decrease in odds of having larger (more negative) differences in Hamaguchi score. As shown in Figure 2B, in both treatment sequences, patients displayed more negative Hamaguchi score changes during the second treatment period as compared to the first one (AB:B vs. AB:A and BA:A vs. BA:B) Figure 2B. In addition, patients receiving Epatrex treatment exhibited on average more negative differences in Hamaguchi score as compared to patients receiving placebo both in the first (AB:A vs. BA:B) Figure 2B and second treatment period (BA:A vs. AB:B) Figure 2B.

Figure 2: Boxplots of Hamaguchi score changes from baseline stratified. Note: A) reatment and WC categories and B) sequence: Treatment period. Boxplots were obtained using observations from all 67 patients. Values inside the boxes indicate mean values. A=Epatrex treatment; B=Placebo.

According to these results, Hamaguchi score changes are not significantly affected neither by age (p=0.9037) or gender (p=0.4103). Instead, a significant impact on the analyzed endpoint was found with respect to alcohol consumption (p=0.0063), with larger Hamaguchi score changes more likely to occur in alcohol consumer patients (odds ratio=0.57 when comparing non-alcohol consumers vs. alcohol consumers).

Discussion

In the present work, we used data from the REVEAL clinical study to investigate the effects of Epatrex treatment on the Hamaguchi score endpoint. Previous analyses of the REVEAL study have been focused on the evaluation of treatment effects on serum markers of hepatic inflammation, such as ALT, AST, GGT and on the Fatty Liver Index (FLI), a biomarker panel derived from BMI, WC, TG and GGT measurements [23]. To date, Epatrex effects have not been evaluated in relation to Ultrasound (US) imaging-based biomarkers, such as the score introduced by Hamaguchi et al. [8] for the assessment of liver steatosis. In the current analyses, although a clear treatment effect could not be identified on Hamaguchi score changes from the available data, a statistically significant relationship was identified between Hamaguchi score changes and WC-treatment interaction, suggesting a stronger effect of Epatrex in patients with high WC values. For patients with normal WC values, effect of Epatrex was less evident. However, it could be argued that the observed differential effects of Epatrex treatment in the two WC categories could be due to the under-representation of the normal WC category with respect to the high one in the analysis population. Indeed, at the first period baseline visit, only about 25% of the enrolled patients belonged to the normal WC category.

Besides assessing treatment effects on the Hamaguchi score endpoint and identifying patient subgroups most likely to respond to Epatrex treatment, a second objective of the present analyses was to investigate the impact of study design and patient-specific factors to provide recommendations for future clinical studies. To achieve this, the Hamaguchi score endpoint was analyzed by means of ordinal multivariate logistic regression, which allowed not only to assess treatment effect but also to uncover study covariates influencing endpoint dynamics independently of treatment. In addition to WC category, factors like endpoint baseline value, period and alcohol consumption were found to significantly affect the endpoint investigated in our analysis. As noted in the European Medicines Agency (EMA) guideline on adjustment for baseline covariates in clinical trials, if a baseline value of a continuous primary outcome measure is available, then this should usually be included as a covariate, since the use of change from baseline with adjustment for baseline is generally more precise than change from baseline without adjustment and it improves the efficiency of the analysis [24].

Concerning the impact of study design factors, we found a significant association between treatment period and Hamaguchi score changes. Although Hamaguchi score changes from baseline were significantly larger in the second treatment period irrespectively of treatment administration, a much larger placebo effect was observed in the second treatment period as compared to the first one most likely compromising the identification of treatment effect statistical significance. Placebo effects may increase over time as a result of the so-called “Hawthrone effect”, by which knowledge to be observed induces patients to adopt a healthier lifestyle behavior [25]. This phenomenon is known to be particularly relevant in NAFLD. In future analyses, data capture over time for diet (caloric intake, dietary composition, alcohol consumption) and physical activity will be considered to adjust for factors recognized to possibly affect placebo response.

In our analyses, alcohol consumption assessed at screening was found to be statistically significant on Hamaguchi score changes from baselines, pointing at alcohol consumption as confounding factor in treatment effect detection. Besides standard questionnaires on alcohol use, future studies should consider longitudinal tracking of biomarkers, such as urinary ethyl glucuronide and phosphatidylethanol, that could allow for a more accurate assessment of influence of alcohol intake during study [25].

Broadly speaking, mathematical models in life sciences can be classified into two categories: Empirical models and mechanistic models. The former category includes statistical models relating a response variable to a number of input covariates. While these models are useful to extract information about treatment effect on a trial endpoint and to inspect the impact of study design and patient specific factors, mechanistic models recapitulating the biology of the phenomena under study may allow extrapolation beyond the range of the data and simulation to answer “what-if” questions [19]. In this study, we opted for a statistical modeling approach, as construction of a mechanistic model would have required more intensive longitudinal data capture to track disease progression and quantify treatment effect over a time.

Conclusion

The outcomes of these analyses support Epatrex beneficial effects in NAFLD patients, as suggested by the results obtained from the Hamaguchi score assessments, which showed sensitivity to treatment period, alcohol consumption and waist circumference.

These findings provided insights on how to ameliorate study design in this patient population and will serve as basis for the development of a mechanistic computational tool to quantitatively optimize Epatrex treatment and clinical trial design in NAFLD patients.

Funding

The present work was funded by Biofarma SRL.

Competing Interests

Vanelli A and Murzilli S are employees of Biofarma, which funded the present work. Bursi R, Nicolò C and Baretta A are employees of InSilicoTrials Technologies, which received funding to perform this work.

References

1. Vernon G, Baranova A, Younossi ZM (2011) Systematic review: The epidemiology and natural history of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis in adults. Aliment Pharmacol Ther 34: 274-285.

   [Crossref] [Google Scholar] [Indexed]

2. Oh MK, Winn J, Poordad F (2008) Review article: Diagnosis and treatment of non-alcoholic fatty liver disease. Aliment Pharmacol Ther 28: 503-522.

   [Crossref] [Google Scholar]

3. Younossi ZM (2008) Review article: Current management of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis. Aliment Pharmacol Ther 28: 2-12.

   [Crossref] [Google Scholar] [Indexed]

4. Neuschwander-Tetri BA, Caldwell SH (2003) Nonalcoholic steatohepatitis: Summary of an AASLD single topic conference. Hepatology 37: 1202-1219.

   [Crossref] [Google Scholar] [Indexed]

5. Ong JP, Pitts A, Younossi ZM (2008) Increased overall mortality and liver-related mortality in non-alcoholic fatty liver disease. J Hepatol 49: 608-612.

   [Crossref] [Google Scholar] [Indexed]

6. Hashimoto E, Tokushige K (2011) Prevalence, gender, ethnic variations and prognosis of NASH. J Gastroenterol 46: 63-69.

   [Crossref] [Google Scholar] [Indexed]

7. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, et al. (2018) The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the study of liver diseases. Hepatology 67: 328-357.

   [Crossref] [Google Scholar] [Indexed]

8. Hamaguchi M, Takao K, Yoshito I, Yuichi H, Kota F, et al. (2007) The severity of ultrasonographic findings in nonalcoholic fatty liver disease reflects the metabolic syndrome and visceral fat accumulation. Am J Gastroenterol 102: 2708-2715.

   [Crossref] [Google Scholar] [Indexed]

9. KozÅ?owska-Petriczko K, Wunsch E, Petriczko J, Syn WK Milkiewicz P (2021) Diagnostic accuracy of non-imaging and ultrasound-based assessment of hepatic steatosis using Controlled Attenuation Parameter (CAP) as reference. J Clin Med 10: 1507.

   [Crossref] [Google Scholar] [Indexed]

10. Ratziu V, Bellentani S, Cortez-Pinto H, Day C, Marchesini G (2010) A position statement on NAFLD/NASH based on the EASL 2009 special conference. J Hepatol 53: 372-384.

    [Crossref] [Google Scholar] [Indexed]

11. European Association for the Study of the Liver, European Association for the Study of Diabetes, European Association for the Study of Obesity (2016) EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease. J Hepatol 64: 1388-1402.

    [Crossref] [Google Scholar] [Indexed]

12. Chalasani N, Younossi Z, Lavine JE, Diehl AM, Brunt EM, et al. (2012) The diagnosis and management of non-alcoholic fatty liver disease: Practice guideline by the American Gastroenterological Association, American Association for the study of liver diseases and American college of gastroenterology. Hepatology 55: 2005-2023.

    [Crossref] [Google Scholar] [Indexed]

13. Lavine JE, Schwimmer JB, Natta MLV, Molleston JP, Murray KF, et al. (2011) Effect of vitamin E or metformin for treatment of nonalcoholic fatty liver disease in children and adolescents: The TONIC randomized controlled trial. JAMA 305: 1659-1668.

    [Crossref] [Google Scholar] [Indexed]

14. Sanyal AJ, Chalasani N, Kowdley KV, Mccullough A, Diehl AM, et al. (2010) Pioglitazone, vitamin E or placebo for nonalcoholic steatohepatitis. N Engl J Med 362: 1675-1685.

    [Crossref] [Google Scholar] [Indexed]

15. Ni Y, Zhuge F, Nagashimada M, Ota T (2016) Novel action of carotenoids on non-alcoholic fatty liver disease: Macrophage polarization and liver homeostasis. Nutrients 8: 391.

    [Crossref] [Google Scholar] [Indexed]

16. ClinicalTrials.gov.

     

17. Statistical Consultancy Team (2018) Nutraceutical trials: Demand, design and challenges.

     

18. Marshall S, Madabushi R, Manolis E, Krudys K, Staab A, et al. (2019) Model-informed drug discovery and development: Current industry good practice and regulatory expectations and future perspectives. CPT Pharmacometrics Syst Pharmacol 8: 87-96.

    [Crossref] [Google Scholar] [Indexed]

19. Nicolò C, Sips F, Vaghi C, Baretta A, Carbone V, et al. (2023) Accelerating digitalization in healthcare with the InSilicoTrials cloud-based platform: Four use cases. Ann Biomed Eng 51: 125-136.

    [Crossref] [Google Scholar] [Indexed]

20. Bellentani S, Murzilli S, Vanelli A (2020) A new combination of 2 nutraceuticals and 1 probiotic (Epatrex) for Non-Alcoholic Fatty Liver Disease (NAFLD): Results from a multi-centre, randomized, single-blind, parallel, cross-over study: The ‘REVEAL’ study. Fatty Liver Disease 3: 13.

    [Google Scholar]

21. R Core Team (2021) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

     

22. Ross R, Neeland IJ, Yamashita S, Shai I, Seidell J, et al. (2020) Waist circumference as a vital sign in clinical practice: A Consensus Statement from the IAS and ICCR working group on visceral obesity. Nat Rev Endocrinol 16: 177-189.

    [Crossref] [Google Scholar] [Indexed]

23. Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, et al. (2006) The fatty liver index: A simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol 6: 33.

    [Crossref] [Google Scholar] [Indexed]

24. European Medicines Agency (2013) Adjustment for baseline covariates in clinical trials: Scientific guideline.

     

25. Rinella ME, Tacke F, Sanyal AJ, Anstee QM (2019) Report on the AASLD/EASL joint workshop on clinical trial endpoints in NAFLD. Hepatology 70: 1424-1436.

    [Crossref] [Goog le Scholar] [Indexed]