Mike Kalutkiewicz

Incredible milestone for Resoundant and Mayo Clinic! The FDA has accepted our LOI into the Biomarker Qualification Program for Liver Stiffness Measurement by MRE for development as a reasonably likely surrogate endpoint (RLSE) in MASH clinical trials. We want to sincerely thank the FDA staff for their thoughtful engagement and guidance, as well as the researchers, clinicians, and study teams around the world whose work over many years has helped build the robust evidence base supporting MRE! Collectively, these efforts now span thousands of patients, enabling rigorous evaluation of fibrosis burden and its relationship to outcomes. Fibrosis remains the key driver of adverse outcomes in MASH, and there is growing alignment across regulators, academia, and industry around the need for non-invasive, quantitative, and scalable tools that can support more efficient and patient-friendly trials.

🚨🚨 Did you catch it? Lilly hinted on their last earnings call that their Phase 3 MASLD trial for tirzepatide and retatrutide may be biopsy-free—using only non-invasive tests like MRE and MRI-PDFF. This could be a watershed moment for metabolic liver disease trials: faster, cheaper, more accurate, and better for patients. The shift from biopsy to NITs in pivotal trials might not be years away - it may already be here.

19% Most imaging biomarkers try to prove they’re specific. MRE did the opposite early on - RSNA QIBA experts were working to standardize MRE for *clinical use* first - not just for clinical trials. So the widely cited 19% repeatability coefficient (RC) for MRE wasn’t cherry-picked to look precise. It was designed by QIBA to reflect the worst-case clinical variation (e.g., patient moves to another state before follow-up, is scanned on a different scanner model, software, didn't fast, possibly with an untrained MRE reader). So the MRE repeatability studies that QIBA pulled from had: - Different scanners, sequences, and software - No fasting requirements - Scan intervals up to 30 days - Different operators and software - Even untrained readers(!) These studies together found a within-subject coefficient of variation (wCV) of 7.9% - which then plussed up according to the Repeatability Coefficient formula of RC = 1.96 x √2 x 7.9% = 19%. Voilà! (side note: the RC formula was created to take into the longitudinal variability described above) So even in the messiest environments, if MRE changes ≥19%, you can be 95% confident it's a true biological change. But here’s what many miss: the 95% confidence interval for the 19% RC is roughly 15% to 24%. That means even QIBA’s “conservative” number has some flexibility built in - if the change in MRE is ≥15% (the lower bound of the 95% CI) you can still be 95% certain. But let's switch to clinical trials. In high-profile Phase 2b or 3 studies, we’re not operating in worst-case clinical settings that QIBA took into account. Here we have very tightly controlled settings: - patients usually receive baseline and f/u on the same scanner - no software updates are allowed during the study - expert readers (or our automated software at Mayo) - strict fasting Fortunately, there have been a few well-controlled studies of MRE in similar environments that yielded wCV of 3.8%. That yields an RC of: RC (95%) = 1.96 x √2 x 3.8 = *10.5%* (CI: 8.5% to 14.5%) So in tightly-controlled clinical trial environments, a ≥8.5% change in MRE represents a true biological change with 95% confidence. But you don't have to be locked into the 95% confidence, which is the same as 5% false positives, or 95% specificity. There are virtually no tests that have 95% spec in any context. So what about 90% specificity, aka 90% confidence? That would be RC (90%) = 1.44 x √2 x 3.8 RC at 90% confidence: 8.8% (CI: 7.2% to 11.2%) That means if FDA allows 90% specificity with your monitoring biomarker, a change of MRE of just ≥7.2% would be significant with 90% confidence. I'd love to see resmetirom data re-run with the proportion of patients who experienced a MRE change of ≥8kPa! I bet the data would be even more impressive! And if you're designing a trial, it's critical to understand that 19% change in MRE is probably too conservative given the efforts you've taken to reduce variability.

A new study in Nature caught my attention for its innovative approach to a long-standing challenge: noninvasively assessing liver inflammation in MASH. lnkd.in/evmimxGB The team (Vanderbilt Radiology; Xiaoyu Jiang, John Gore et al) introduces MR cytometry - an imaging technique that estimates cell size and density using MRI data. The rationale: 🟢 In healthy livers, large hepatocytes dominate. 🟠 In MASH, inflammatory processes bring in smaller immune cells (like lymphocytes and macrophages), increasing cell density and decreasing average cell size. By setting thresholds (cell size <10 μm and density >30 x 10⁴/mm³) the authors quantify inflammation-related changes across the liver. The key result (Fig. 7) is below - and I've never seen such clean stratification using any biomarker (serum, imaging, etc) in MASH. MASH patients had a significantly higher percentage of liver voxels matching this inflammatory signature compared to healthy controls - measured noninvasively on practically any standard 1.5T or 3T MRI scanner, in under 12 minutes. Because this is a direct method for assessing cellular characteristics of the liver, MRI cytometry would be a very welcome tool for clinical trials in terms of quantifying inflammatory reductions alone. "This technique holds promise as a complementary tool to existing noninvasive methods, such as PDFF and MRE, contributing to the development of a comprehensive, noninvasive diagnostic approach for MASH." It's still early work, but this is worth following closely.

"In clinical guidelines, we'll advocate for MRI-PDFF for MASH assessment and VCTE for fibrosis assessment." "Sir, why not do MRE simultaneously with PDFF, since it only takes a minute longer, is more accurate, and actually makes the exam cheaper?" "Nobody knows." #livertwitter

Funny moment this morning watching my kids get ready for school, and then mom comes on TV talking about tariffs/trade 🙂

Happy to be co-author on a great and highly practical study (lead author Yasaman Vali) from the LITMUS Consortium on the clinical factors that might require adjusted thresholds to noninvasive tests (NITs) for liver fibrosis assessment. Some key takeaways: 🟢 NITs like FIB-4, ELF, and VCTE can have a high accuracy for ruling-in and ruling-out liver fibrosis in MASLD/MASH patients. 🟠 However, clinical factors such as BMI can have a major impact on the thresholds required to maintain this accuracy, particularly for VCTE. The kPa threshold to maintain 90% specificity (thus 10% false positives) needs to increase from 10kPa to 15kPa as BMI rises from ≤25 to ≥30 (purple circle). This is a 50% change in the threshold, compared to a modest change for ELF across the same patient cohort. Why is this important? The BMI in the pivotal MAESTRO-NASH trial was 35, meaning that for most MASLD/MASH patients, a threshold of 15kPa or greater will be needed rule in advanced fibrosis with only 10% false positives. Without adjustment (i.e., using a threshold of 8 kPa for a patient with a BMI of 32), you will greatly increase the number of false positives, resulting in healthy patients being told they have disease and undergoing unnecessary treatments. 🟠 The thresholds for accurately ruling-in (90% sensitivity) and ruling-out (90% specificity) at a BMI >30 are basically 8-15 kPa, which ironically is the same range used by a number of professional organizations for ruling-in for new resmetirom treatment. The problem? This is essentially an indeterminate zone for BMI ≥30, rather than a zone to confidently rule-in disease. 🧠 Key takeaway: more work is needed to better define the use of these tools across different clinical profiles. VCTE in particular seems highly sensitive to elevated BMI and may need custom/adjusted thresholds based on BMI. ELF doesn't appear to suffer as much from this confounding factor. lnkd.in/e7r2D5Fv #livertwitter #MASLD #MASH

Interesting paper here. Couple of quick thoughts: - a bad test (low sensitivity) is the worst thing. Patients value most: "if I have the disease, I want the test to find it early" - False Positives: the data here seem odd. As the FP% decreases from 40% to 20% and then 10%, the utility goes down? I think participants didn't quite understand the concept? Maybe the questions should have linked False Positives to their real-world consequences: anxiety, unnecessary treatments, unnecessary side effects, etc. - Cost, time and type of test are all pretty marginal in comparison to a test's accuracy - abbreviated MRI (aMRI) did extremely well due to its optimal sensitivity