Abstract

Background: FDA recommended histopathological end-points for NASH clinical trials may not be fully evaluated by conventional histological staging. Shorter time frames in NASH trials make it difficult to assess changes in treatment-induced fibrosis with current classification systems. Assessment of specific morphological fibrosis parameters and their quantification in specific zones of NASH-CRN classification, can be done by SHG/TPE-based AI platforms. NASH-CRN parameters may map overall fibrosis changes, but any specific drug-induced changes may not be fully quantified by these alone. This detailed analysis helps overcome limitations of conventional histological examination and describe fibrosis changes and drug-induced effects.

Methods: Biopsy slides were obtained from FALCON-1 clinical trial (24-week randomized double-blind, placebo-controlled phase 2 study of 10, 20, and 40 mg pegbelfermin in NASH patients, NCT03486899). SHG/TPE microscopy and AI analysis were used to estimate NASH-CRN fibrosis parameters as continuous variables for 176 paired biopsy slides. Additional Zone 2 fibrosis parameters were evaluated to measure drug-induced effects. Based on pathologist classification of changes in fibrosis stage of biopsies comparing baseline and post-treatment slides, they were divided into three groups: Progressive (1-stage or more increase in fibrosis), Regressive (1-stage or more decrease in fibrosis), and No-change (no change in fibrosis stage). These changes were depicted using radar maps with normalized values.

Results: In regressive group, there was reduction of fibrosis as per NASH-CRN parameters in both treatment and placebo cohort, measured by SHG/TPE microscopy and AI analysis. Reductions in placebo cohort were statistically insignificant; in treatment cohort (which included patients of all drug doses combined), reduction in fibrosis of periportal and perisinusoidal regions were statistically significant (Figure 1A). Fibrosis measurements in No-change group follow similar pattern of reduction as in Regressive group (Figure 1B). Specific Zone 2 fibrosis parameters (S1-S3) for treatment-cohort of all 3 groups are shown in Figure 1C. These show statistically significant reductions in fibrosis parameters post-treatment, despite being labelled as regressive, no-change and progressive by pathologist classifications. Parameters S1-S3 in placebo-cohort also show reductions, but were statistically insignificant.

Conclusion: SHG/TPE-based AI platforms allowed for quantification and mapping of specific parameters revealing improvement in both NASH-CRN and Zone 2 fibrosis parameters not seen in manual NASH-CRN staging. They may be able provide quantitative drug-induced specific changes in fibrosis which can be more sensitive than manual staging. This may be ideal for phase 2 clinical trials which have short time frames, and help better design and strategize phase 3 trials.