Year: 2009
Type: article
Abstract: In addition to work being done to characterize the geoepidemiology of nonalcoholic fatty liver disease (NAFLD),1 there have been equally important discoveries surrounding the pathogenesis of this liver disease. At one time, a condition of "two hits" was thought to be necessary for the origin of this disease, but there is now evidence pointing toward several additional mechanisms contributing to the liver injury known as nonalcoholic steatohepatitis (NASH).2 NASH remains an important phenotypic state, because this subgroup of patients is deemed at high-risk for developing progressive disease resulting in cirrhosis, liver failure requiring transplantation, or death.3 CK-18, cytokeratin 18; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; ROC, receiver operating characteristic. Liver histology remains the gold standard for identifying patients with NASH. Although much has been said about the sampling error and observer variation with liver biopsy,4 there are also concerns regarding increased resource utilization and reduced patient acceptance which will further limit the population-wide use of this test for detecting NASH.5 Because effective therapies are being developed and tested for NASH, questions will remain as to whether fibrosis assessment with liver biopsy is sensitive enough to detect small but potentially important changes over time. Given the preponderance of individuals being identified with NAFLD, the recognition of a noninvasive, versatile test to discriminate patients with NASH from fatty liver alone would be highly practical and useful. In this issue of HEPATOLOGY, Feldstein et al.6 present results from a study examining the role of plasma cytokeratin 18 (CK-18) fragments as a biomarker distinguishing patients with NASH from those with fatty liver alone. The rationale for this study was based on emerging data suggesting that hepatocyte cell death by apoptosis is typically present in animal models and humans with NASH,7 including preliminary studies implying a potential diagnostic role for CK-18 in detecting NASH.8-10 The study population consisted of 139 subjects with biopsy-proven NAFLD representing the National Institutes of Health (NIH) sponsored NASH Clinical Research Network and 150 age-matched controls. Plasma CK-18 was measured for each subject with NASH using a blood sample taken within 3 months of the liver biopsy obtained from the NIH blood bank repository. Plasma CK-18 measurements were collected and processed in different centers using the same enzyme-linked immunosorbent assay, while the type of histological injury from NAFLD was verified by review from a single hepatopathologist. The results were noted for significantly higher CK-18 fragment levels in subjects with NAFLD as compared to the 150 healthy volunteers. In addition, the average values were significantly higher in patients with NASH versus those without NASH or a borderline diagnosis of NASH. However, the CK-18 levels could not distinguish between subjects with NASH or an indeterminate diagnosis of NASH. A modest yet positive association was also noted between mean plasma CK-18 values and histological activity score and stage of hepatic fibrosis. Finally, after adjusting for potential confounding variables, plasma CK-18 remained an independent predictor for the presence of NASH. As a diagnostic test, the CK-18 fragments retained a sensitivity between 65%–77% and specificity between 65%–92% for identifying NASH based on various diagnostic threshold values. The advantages of this study include (1) its use of a multicenter approach, (2) the availability of well-characterized individuals with NASH, and (3) the study of a plasma-based apoptosis marker which appears relevant in the pathogenesis of NAFLD. Potential limitations include (1) the absence of reported imaging studies or liver biopsies in healthy control subjects to exclude NAFLD, (2) the selection of controls from a single center which may not have been representative of potential controls selected at other centers, (3) the absence of data on body mass index for controls and the possibility that CK-18 values could be influenced by obesity/insulin resistance in the absence of NAFLD, and (4) data on liver histology specimen size and portal tract number to determine the potential effect of sampling variability in recognizing NASH. Several points related to study design are worthy of discussion. The frequency of histological findings consistent with NASH was observed in more than 50% of subjects within this study. In turn, a more even distribution of histological findings would be highly desirable in future studies to minimize the potential for spectrum bias on test performance results. We also know sampling error in NAFLD can be significant. As stated by the authors, utilizing liver biopsy alone for risk-stratifying individuals with a common disorder will result in unwanted misclassification of disease stage. However, plasma CK-18 fragments representing hepatocyte apoptosis may also be subject to potential sampling variability when assessing its role as a diagnostic test. As with all new diagnostic tests, there are several other caveats to keep in mind. Understanding the reproducibility of plasma CK-18 fragment values will be required for additional investigations of this test. Knowledge regarding the variation in measurement over time and between centers will be critical, should CK-18 be used as a biomarker for assessing therapeutic effect in subsequent, multicenter clinical trials. Whereas the calculation of mean CK-18 fragment values is helpful to perform statistical tests between patient subgroups for identifying relevant differences, this approach will be less helpful when assessing the diagnostic performance of CK-18 fragments. Clinical experience tells us that tests which provide continuous values still require us to identify an optimal threshold (or cutoff) for assessing the likelihood that a condition or disease is present (such as white blood cell count in a patient with suspected appendicitis). Assessing the sensitivity, specificity, positive predictive value, and negative predictive value are valuable in understanding the potential performance of a predictive laboratory test. In this study, one can observe that the area under the curve value for CK-18 fragments was considered excellent at 0.83. Subsequently, the sensitivity (71%–75%) and specificity (81%–85%) for certain plasma CK-18 fragment levels correspond to optimal areas of the receiver operating characteristic (ROC) curve. These results, however, suggest the possibility that false-positive test rates between 25%–29% and false negative rates of 15%–19% could occur at these corresponding diagnostic threshold values. Furthermore, these results might be considered ideal since the procurement of satisfactory liver specimens and measurable CK-18 levels for individuals in clinical practice will not be as complete when compared to experimental settings. The impact of these misclassification rates, as well as patient preference and long-term outcomes resulting from testing, will also require further study in determining the clinical effectiveness of this approach for identifying patients with NASH.11 However, as these and other biomarkers become more sophisticated, they are likely to take on an increasingly important role in the diagnosis and management of NAFLD. Finally, the validation of a noninvasive biomarker for NASH (or any liver disease) requires additional work beyond verifying diagnostic test performance characteristics among independent populations. From a regulatory perspective, the validation of a biomarker usually refers to the assay method under investigation, while qualification of a biomarker defines the evidentiary process connecting the biomarker, the biological process and the clinical end point together.12 Organizing efforts to "qualify" biomarkers for NASH and other liver disease would require collaboration within consortia in cooperation with the Food and Drug Administration. This formalized, central process was designed to overcome the complexity and lack of direction provided by unstructured processes for biomarker approval to date.13 The viability of this approach, however, remains somewhat uncertain but for now is the pathway necessary to incorporate desired biomarkers into clinical trials of novel drug therapies.14 Despite the complexities of biomarker validation, there have already been several non-invasive diagnostic indices presented in the literature for detecting clinically significant fibrosis from NAFLD.15 Although biomarkers for NAFLD have also focused on mechanisms of oxidative stress, inflammation, and apoptosis, the majority of recent data continues to revolve around fibrosis detection. Recent investigations of multivariable panels have demonstrated improved diagnostic performance over single marker indices even when studied among independent populations.16-19 (Table 1). Elastography imaging with transient ultrasound20 or magnetic resonance21 platforms also provides excellent diagnostic accuracy for detecting advanced fibrosis in human subjects with NAFLD. For both serum and elastography imaging tests, sensitivity and specificity values have exceeded 80%, respectively, in detecting advanced fibrosis when compared to liver biopsy. Additionally, negative predictive values of 95% or greater help to improve our confidence in excluding the presence of advanced disease. However, the data from these studies remains cross-sectional in nature, and persistent issues of spectrum bias and variable histologic accuracy as a gold standard continue to prevent direct comparisons between these studies to determine which noninvasive test should move forward to clinical effectiveness trials. Furthermore, the ability to distinguish individuals with early stage NASH (i.e., inflammation and no fibrosis) from those with simple steatosis alone is as clinically relevant as detecting patients with advanced fibrosis. Although this scenario has not been often addressed in noninvasive studies to date, emerging data in animal models verifies preliminary human experiences21 that magnetic resonance elastography may have the potential to serve this purpose in the future.22 This study by Feldstein and colleagues demonstrates the simplicity, biological plausibility, and potential widespread availability of plasma CK-18 for improving our ability to risk-stratify patients with NAFLD. It would also appear to provide an advantage for use in proof-of-concept clinical trial settings examining novel therapies geared toward inhibiting apoptosis, but additional work is required before we can know how it will perform in the clinic. One possibility is that CK-18 fragment levels will function as an "added value" test which defines algorithms for subsequent testing in patients to confirm or deny the presence of NASH. Whatever the optimal diagnostic approach or approaches will be, information on the comparative effectiveness of these strategies will be crucial for determining how we as clinicians risk-stratify our patients with NAFLD for therapeutic interventions in the future. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
Source: Hepatology
DOI: 10.1002/hep.23288
Author: Jayant A. Talwalkar
Institution: Mayo Clinic
Language: English
Subfield: Epidemiology
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W2001537101