Reducing NAFLD-Screening Time: A Comparative Study of Eight Diagnostic Methods Offering an Alternative to Ultrasound Scans

Filippo Procino; Giovanni Misciagna; Nicola Veronese; Maria G. Caruso; Marisa Chiloiro; Anna M. Cisternino; Maria Notarnicola; Caterina Bonfiglio; Irene Bruno; Claudia Buongiorno; Angelo Campanella; Valentina Deflorio; Isabella Franco; Rocco Guerra; Carla M. Leone; Antonella Mirizzi; Alessandro Nitti; Alberto R. Osella; MICOL GROUP

Disclosures

Liver International. 2019;39(1):187-196. 

In This Article

Discussion

In this work, a high NAFLD prevalence has been shown, affecting a higher proportion of males than females. The NAFLD group had significantly higher anthropometric and serum biomarkers values revealing metabolic impairments than the NO-NAFLD group. The values of the eight formulas analysed resulted higher in the NAFLD than in the NO-NAFLD group. Their predictive power resulted good, but better in females than in males. Adjusting for age elicited slightly better AUROC values for all anthro-m (except AVI) only in males. Regarding the performance of each formula employed in a two-step hybrid method for NAFLD diagnosis in a large population, WHtR showed the best performance in reducing the need for US, followed by HIS and FLI. Considering the false negative rate (NAFLD missed %), the best performance was achieved by AVI while WHtR was the worst. The best percentage of identified NAFLD was obtained using AVI.

There is no consensus about the prevalence of NAFLD. The actual prevalence is possibly underestimated because a significant proportion of the patients are asymptomatic, present slight biological alterations and so do not undergo appropriate instrumental tests or even any tests. Depending on which diagnostic method is used, the prevalence may result different. The best test to diagnose NAFLD, in order of accuracy, is the liver biopsy or MRI but, because of the difficulty and cost, US is found preferable, especially for large population screening as in our study.[34] We detected a 31.6% prevalence of NAFLD, higher than the global prevalence (24%) but similar to previous studies in European populations, like the SHIP Study, that estimated the prevalence of NAFLD to be ~30% when diagnosed by ultrasonography, or like a postmortem study conducted in Greece, that revealed simple steatosis in 31.3%.[35,36]

Our findings confirm significant differences in the critical markers of metabolic impairments in the NAFLD population vs NO-NAFLD patients. In both sexes, we observed that the BMI was above the obesity cut-off in the NAFLD group, vs the overweight area in the NO-NAFLD group. This outcome is similar to findings in other studies, in which a higher prevalence of NAFLD was detected in obese vs overweight or normal weight subjects.[15]

Our results show higher values of parameters making up the metabolic syndrome frame. In the NAFLD group, there were significantly higher levels of WC and TG vs the NO-NAFLD group in both sexes, confirming findings in other studies showing that the most subjects with NAFLD had many features of the metabolic syndrome or at least one of the Adult Treatment Panel III (ATP III) criteria.[11–13]

We also observed significantly higher levels of hepatic enzymes (ALT/AST and GGT) in the NAFLD vs NOT NAFLD group, although the measured mean in the NAFLD group was not outside the normal range for the healthy population in either sex. These findings confirm the risk of underestimating the true NAFLD prevalence and are in agreement with results from other research that reported elevated ALT serum levels in only 54% of NAFLD cases.[11]

The values of the eight formulas analysed (FLI, HIS, BMI, WC, AVI, WHtR, WHt_5R and BRI) also resulted significantly higher in the NAFLD vs NOT NAFLD group in both sexes confirming the possibility of using these formulas for identifying the NAFLD risk.[17–23,26]

Our findings confirm a good predictive power of all the formulas analysed, compared using AUROC. The widest AUROC, in both sexes, was for FLI, which confirms the indications in the EASL-EASD-EASO guidelines.[17] Although it was only validated in the Korean population, the second widest AUROC was for HIS, confirming the usability of this formula also in our population.[26] Anthro-m also showed a good performance, especially for WC, AVI and WHt_5R.

WHtR, BRI and BMI registered the smallest values of AUROC, although they were still ample enough to yield a good performance (0.82 for females and 0.74 for males). These findings confirm the predominant role of metabolic aspects in NAFLD and the applicability of anthro-m for NAFLD prediction, as already described in other studies in non-European populations.[19,20]

All formulas had a wider AUROC for females than in males. For all anthro-m, except AVI, adjusting for age slightly improved the AUROC values in males but not in females, while for FLI, HIS and AVI adjusting for age did not improve the AUROC values either in males or in females.

Considering the different influence of sex and age on the AUROC, we believe it is more appropriate to calculate the optimal cut-offs for males and females separately, and considering age only for the BMI, WC, WHtR, WHt_5R and BRI formulas in the male group.

Applying the cut-offs obtained, we compared the real performance of each formula. The results, shown in Table 5, suggest that all the formulas can reduce the number of US. The best performance was obtained with WHtR, HIS and FLI, and the worst performance with WC and AVI. These findings are in accordance with the PPV of each formula considered. On the other hand, considering the NAFLD missed %, WHtR yielded the worst and AVI the best performance. These values are congruent with NPV values. Due to the reduced need for US, obtained using the formulas, the prevalence, in the group of subjects at risk, increases from 31.6% to 47.8%-51.9%. Nevertheless, the percentage of subjects was not sufficiently reduced to indicate passing directly on to the therapy step.

Analysing the percentage of identified NAFLD over the total affected subjects, we obtained indications as to the real performance of the hybrid alternative diagnostic methods. The best percentage of NAFLD identified was obtained with AVI. Considering the smaller false negative rate, AVI resulted the best formula to combine with US in diagnosing NAFLD and seems more useful than others, because it does not need to be coupled with any biochemical assay. These results are relative to the prevalence of NAFLD registered in our population (31.6%). Figure 3 shows the trend of PPV and NPV vs prevalence and confirms that the findings are repeatable and confirmed at different prevalences of NAFLD.

In conclusion, gender was the major factor influencing the AUROC, in order to identify the optimal cut-offs of all the formulas considered, while age only affected the anthro-m (excepted AVI) in males.

FLI resulted the best formula to identify people with NAFLD in one step. This formula combined a high PPV and NPV. Nevertheless, the percentage of NAFLD subjects identified as at risk by using FLI (48.2%) was insufficient to prescribe any therapy as more than 50% of subjects were false positive. As a consequence, US needs to be always considered in the screening process of NAFLD diagnosis.

Our findings suggest that the best formula to use in order to obtain the maximum percentage performance of the hybrid method for NAFLD screening in large numbers of subjects was AVI that combined a good reduction in the need for US (although not the best), with the smallest false negative rate (% NAFLD missed). Considering a 31.6% prevalence of NAFLD, the optimal cut-off for AVI was 15.14 L for females and 17.67 L for males. The trend of NPV vs prevalence confirms that AVI remains the best of the all formulas analysed at the prevalence ranging between 20% and 40%. Furthermore, it is simple to apply, because it requires only two anthropometric measures (WC and HC). Considering that AVI was not the best formula concerning reducing the need for US, other studies investigating the advantageous economic aspects are needed to individuate the most convenient way to perform large-scale screening studies for NAFLD.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....