S2-E62 - Mazen Noureddin, Alina Allen And Wayne Eskridge Join SurfingNASH's 2021 NAFLD Year-in-Review

Send us Fan MailThe last half of December marks our annual NAFLD Year-in-Review. Episodes 62-64 each include ~20 minute segments of longer interviews with Stakeholders who have made a dent in Fatty Liver disease in 2021. In this episode, Louise Campbell and Roger Green are joined by Mazen Noureddin,  Alina Allen, Wayne Eskridge.Highlights: 7:51 – Mazen Noureddin begins talking about AI and histopathology by pointing to recent review papers listeners should read, one from Mayo Clinic and another from Mazen and colleagues. 8:21 – AI tools: machine learning and deep learning. Key AI targets: non-invasive testing and histology. More recent progress in histology.10:09 – History of why histology in NASH has become increasingly controversial over time and how the challenge will grow. AI can address many of these controversies.12:42 – Paper earlier this year: three expert pathologists working in tandem can strengthen reliability AND their results correlated with PathAI. AI also provided more granularity on specific disease levels within a cirrhotic liver and reduced placebo success rate.17:10 – Louise: Should Intercept use AI methods in their re-read of 18-month biopsy data? Mazen: yes. We need a path to get regulators to accept AI this way.19:40 – Louise: Why don’t we use Ishak instead of Metavir for cirrhosis coding? Mazen:  another great question!  22:01 – Alina Allen begins23:20 –  In the past year, we have learned that not only is MRE the most accurate estimator of fibrosis, but we can diagnose now based on liver stiffness (LSM) + PDFF, all in an automated 5-minute process.24:51 – Mayo CGH study (2021) proves we can correlate LSM today with five-year outcomes. Provides far better trial screening and matching data than simple biopsy. 27:07 – Roger: how can this change how we coach and advise patients?  29:04 – Alina: we can use LSMs today to inform patients better, improve their care and allocate use of expensive tests more properly.  30:55 – Roger: how does this fit into a Critical Care Pathways world. Alina: we can improve 1st-line granularity via AI analysis of patient’s medical history. If this filters more people out, it makes elastography or a blood-based biomarker more efficient, therefore more affordable,.33:07 – Roger notes that sounds more like common commercial uses of AI than the AI-based histology reading we discussed with Mazen.   35:43 – Louise: if you had a magic wand, what model would you create to make MRE available when appropriate. Alina – first, teach people it is not too expensive or  difficult, then create regional centers of excellence by putting software on existing MRI machines. Finally, educate patients and physicians. 38:10 – Alina: you do not need a full 45-minute MRI to get an MRE, but only a  5-10 minute process. thereby saving time and money.41:03 – Wayne Eskridge begins42:35 – Wayne: decision to start the SUNN study borne of dissatisfaction over recommended standard NOT to screen for NAFLD. 43:35 – Study pitch received unexpectedly positive reception from industry but learning about IRBs and trial management was a “real education.” 46:12 – Wayne: point of study was to go outside medical system to asymptomatic, uninformed people, in the belief they had underlying health concerns and would tbe motivated to learn.47:48 – Louise: "Amazing" study identified potential clinical trials candidates.  Why does SUNN show far higher levels of fibrosis than the other studies?48:56 – Wayne: this is not a general population study, numbers probably are higher. 52:19 – Peak age of 40-50 is younger than with most studies; we know it progresses steadily. Message: we should be screening high-risk patients younger.55:14 – Wayne: Hard cost was ~$120. Took 8 months to complete running heavy patient loads.