I want to raise an underappreciated issue: NIT performance during GLP-1 therapy may differ from baseline performance. Most NIT validation studies were done in treatment-naive patients. When we use these tests to monitor treatment response, we're extrapolating. There are reasons to think the tests might behave differently during treatment: 1. FibroScan stiffness can decrease due to reduced inflammation, not just reduced fibrosis. A patient whose stiffness drops from 10 to 7 kPa on semaglutide may have resolved inflammation (which happens quickly) without meaningful fibrosis regression (which takes years). We could be over-interpreting early FibroScan improvements. 2. FIB-4 components change on GLP-1 therapy. ALT decreases (lowering FIB-4), but weight loss can also mildly decrease platelet count (raising FIB-4). These opposing effects can make FIB-4 trends unreliable during active treatment. 3. ELF score components (HA, PIIINP, TIMP-1) reflect fibrosis turnover, not just fibrosis quantity. During active fibrosis regression, these markers might paradoxically rise before they fall as the extracellular matrix is remodeled. This is why paired biopsy remains the gold standard for assessing treatment response in clinical trials. For clinical practice, we use NITs because they're practical, but we should acknowledge their limitations during active treatment. The most reliable approach: use multiple NITs together and look for concordant trends. If FibroScan, ELF, and ALT are all improving, you can be confident. If they're discordant, interpret cautiously.

This is an important nuance and I've seen it clinically. I had a patient whose FibroScan dropped dramatically from 11.2 to 7.8 kPa in just 4 months on tirzepatide. I was excited and told her the fibrosis was improving. Then I looked at her ALT — it had dropped from 88 to 32 in that same period. The FibroScan improvement was almost certainly driven primarily by the reduction in necroinflammation (which contributes to stiffness) rather than by actual fibrosis regression. True collagen remodeling takes 12-24 months minimum. I've since become more conservative in how I interpret early FibroScan changes. My rule of thumb: - 0-6 months: FibroScan improvement reflects reduced inflammation and steatosis - 6-12 months: FibroScan improvement likely reflects a mix of reduced inflammation and early fibrosis regression - >12 months: FibroScan improvement more likely reflects true fibrosis regression This is why I don't repeat FibroScan before 6 months anymore, and I consider the 12-month FibroScan the first truly meaningful assessment of fibrosis response. For patients tracking their numbers at home — don't be discouraged if your 6-month FibroScan shows less improvement than you expected. The stiffness may have plateaued after the initial inflammation-driven drop, and real fibrosis improvement takes longer to manifest.

To tie this all together with a practical take-home for trainees and community providers: My NIT decision matrix for MASH patients on GLP-1 therapy: | Clinical Scenario | Primary NIT | Secondary NIT | Frequency | |---|---|---|---| | F0-F1, starting GLP-1 | FIB-4 | FibroScan if available | Every 12 months | | F2, on GLP-1 therapy | FibroScan | FIB-4 + ELF if available | Every 6-12 months | | F3-F4, on GLP-1 +/- other MASH therapy | FibroScan + ELF | Consider MR elastography | Every 6 months | | Discordant results (improving ALT but static FibroScan) | Repeat FibroScan at 3 months | Consider biopsy if persistent discordance >12 months | Case-by-case | | Post-treatment monitoring (off therapy) | FIB-4 | FibroScan annually | Every 6-12 months | The key principle: no single NIT is sufficient. Use at least two complementary tests (one serum-based, one imaging-based) and track trends rather than individual values. As the field matures, I expect we'll see composite NIT scores that integrate multiple biomarkers into a single treatment-response metric. For the board exam: know the FIB-4 calculation, FibroScan cutoffs, and the AASLD NIT algorithm. And know that NITs are validated for diagnosis but have limitations during treatment monitoring — that distinction comes up frequently.