Co-authored by Gavin Nichols.
Nonalcoholic steatohepatitis (NASH) development programs are shifting focus from targeting the clinical impacts of NASH on reversing the effects of fibrosis and cirrhosis, to long-term chronic management of the causes of NAFLD/NASH through reduction in liver fat. This change in therapeutic focus will rely on diagnostics endpoints that have increased frequency, specificity and sensitivity over the existing liver biopsy. In addition, although the prevalence of NAFLD/NASH is extremely high, its asymptomatic nature combined with the current approach to clinical research with biopsies means the process is expensive, long, risky, and uncertain for patients, investigators and sponsors alike.
How can we best deliver the next wave of fat-related NASH studies? Imaging has multiple modalities that can address the issues above.
Clinically, the goal of recent intervention studies for NASH has been to reduce the metabolic stress and liver fat levels before the path of inflammation and fibrosis becomes irreversible. To achieve this, interventions aim to decrease the amount of lipid droplets in the hepatocyte, and imaging is used to measure the effect of the intervention on liver fat. Here, I’ll discuss several considerations in the clinical trial experimental design for NASH studies to help sponsors achieve their goals.
- Engage imaging experts during protocol development
Radiologists and scientific experts can provide the critical background and clinical feedback to choose the appropriate imaging modality. Using imaging to measure the effect on liver fat requires several considerations.
- What level of liver fat is necessary for inclusion?
- How much of a change from baseline is clinically relevant and predicts or reflects a beneficial clinical outcome?
- Are multiple imaging modalities or measures needed to demonstrate that the clinical effect is real? What if the measurements do not reflect the disease outcome?
These questions are not specific to NASH studies, but remain relevant because the scientific and regulatory community have yet to reach consensus on the best answer to each. However, in a review article for which I was a co-author, we surveyed the recent literature and concluded that the amount of baseline hepatic fat is critical for the design of study objectives and reasonable response estimates in NAFLD and NASH trials, particularly because the identification of early signs of efficacy (or the lack thereof) might be critical for planning future larger trials i.e., being able to identify candidates who are likely or unlikely to respond to the intervention.1
- Define how imaging will be used during your trial
The role imaging will play in selecting subjects and generating efficacy endpoints has a direct impact on the most appropriate imaging modality. Consider the following when you are designing your study with imaging endpoints:
Eligibility – Hepatosteatosis can vary significantly in the NAFLD/NASH population. So it is important to carefully select the population for your endpoints. Eligibility should leverage prior biopsy results whenever possible, provided they are within the 24-week window immediately before the study. To facilitate a change in hepatic fat, start with a subject sufficiently burdened with at least an 8-10% hepatic fat fraction. Prescreening for fat and/or fibrosis with a locally read qualitative ultrasound scan may aid enrollment by filtering out ineligible subjects and enriching the population sent to MRI. MRI is often performed at the end of the screening window to minimize the number of scans performed on subjects who screen fail. With MRI timing occurring late in the screening window, results of central analyses of MRI proton density fat fraction (MRI-PDFF) to confirm eligibility must be delivered quickly, typically within 3 days.
Frequency – Preclinical results should drive the frequency of imaging. In most cases, the liver preserves fat stores for a very long time. Therefore, we recommend imaging at 12 weeks, because shorter time frames may not be sensitive to change. Continued dosing to 24 weeks extends the efficacy follow-up period and contributes to the drug safety database. As imaging tests have a higher sensitivity over ultrasound or biopsy, sites can test more often, which may reduce the overall time of the trial, the site/patient burden and overall costs.
Change –Change from baseline (as observed by imaging and correlated to biopsy results) is considered clinically meaningful and is an active area of investigation and validation. For fat fraction using MRI-PDFF methods, obtaining a relative reduction of ≥30% (i.e., from 20% to 14% fat fraction) has been correlated with a histologic improvement in adipose level. For corrected T1 (cT1), a measure of inflammation and early fibrosis, the benchmark for a significant reduction in inflammation is 50 msec. Magnetic resonance elastography (MRE) is able to measure the liver stiffness that is related to fibrosis. While it remains uncertain whether available treatments can reverse fibrosis and thus reduce liver stiffness, MRE can be used to assess current stiffness status and evaluate the slowing of progression in fibrosis. The reproducibility and sensitivity to change of these methods are published and should be used to design your statistical endpoints. As imaging-based tests have a higher specificity than ultrasound or biopsy tests, the time to detect an efficacy signal may be reduced here as well.
Biopsy – The imaging community is working very hard to eliminate the need to perform biopsy in late-stage development. Because of the morbidity associated with repeat liver biopsies, the significant cost, and the burden to patient/site, non-invasive methods should result in greater compliance with study visits and fewer patient drop-outs. The endpoints themselves vary for liver trials and will be driven by your intended patient population.
Fat Fraction – Because the fat deposition in the hepatocyte is microvesicular and much smaller than the size of the MRI voxel, hepatic fat is reported as the fat fraction. We recommend the MRI-PDFF methodology available on many clinical scanners, at either 1.5T or 3.0T. MRI-PDFF adjusts for several confounders to the measurement and provides the most reliable in vivo assessment of fat across the entire liver. MRS is another method to assess liver fat but is rather more complex than MRI and less widely available, making it a poor choice for multi-center clinical trials.
Stiffness – For patients with a fibrotic component to their liver, measuring liver stiffness using MRE is an excellent biomarker, as stiffness correlates with the amount of fibrous tissue seen on biopsy.
Inflammation – Inflammation and early signs of fibrosis can be quantified using cT1 methods. Since edema changes the free/bound water state of the hepatic parenchyma, these T1 mapping methods are applicable for subject populations with NASH or early fibrosis.
Liver Iron – Liver iron is an imaging confounder as it affects the relaxation properties of the liver tissue and can falsely alter measurements of hepatic fat and inflammation. Proper image acquisition and analysis techniques can minimize the impact of liver iron, assuring accurate measurement of the parameters of interest. Certain diseases, e.g., thalassemia, can alter liver iron content and the ability to measure these changes as a study endpoint becomes of interest. T2* mapping methodology allows computation of liver iron concentration.
Adipose Depots – In certain drug targets, major fat depots in the muscle, subcutaneous, visceral, epicardial or pericardial space may change over time, becoming secondary imaging endpoints. Repositioning a subject to capture datasets for these fat depots does not add a significant amount of time to an MRI protocol and may be worth the investment in some NASH/NAFLD studies.
- Utilize experienced sites
Selecting sites with specific liver experience is important to ensure a cost-effective and efficient start-up. There are several manufacturers of MRI systems with multiple field strengths. Beyond having a high field magnet, sites need additional software or hardware to perform the scans required for NASH endpoints:
- MRI-PDFF methods require special software to process the multi-echo data and generate fat, water and fat fraction images;
- MRE requires a combination of software and hardware i.e., the Resoundant acoustical generator, to generate stiffness maps;
- cT1 requires specific liver sequences which must be installed and verified on each scanner.
Each modality has published reproducibility values, and each site should be qualified for study use using a combination of phantom and normal volunteer scans. In addition, you’ll want to determine if the sites have the imaging modality that you would like to use for your endpoints.
- Assess image partner capabilities
Central imaging services are typically recommended for a study with imaging endpoints for safety and efficacy evaluation of new therapies. Because there are many imaging vendors to choose from, make sure to carefully review their strengths and weaknesses for the modalities utilized in the study.
In summary, imaging modalities should be considered a viable part of the current wave of fat related NASH studies.
|Advantages of Using Imaging for NASH Trials|
|Non-invasive and human readable||Improve patient engagement and acceptance of need to be included in a studyLower screen failuresSegment sub-population||Faster recruitmentReduced overall costRicher scientific insights|
|Higher specificity||Reduced CoV and uncertainty for patients, investigators and sponsors||Smaller trial populations|
|Higher sensitivity||Detect smaller signals, quicker||Shorter trial durations|
The remaining strategies of applying imaging standards and guidelines, providing comprehensive training, using efficient and effective imaging processes that are proactive rather than reactive are all also relevant and should be considered for all imaging studies. I encourage you to review my colleague Tim Crow’s blog for more information and make sure that your chosen vendor adheres to these processes.
Consider discussing costs in the context of total cost. Ultrasound is cheaper but its CoV is significant higher than MRI, so the total cost/time/patient burden may be more.