◉ 5-YEAR SURVIVAL CURVES · PANCREATIC DISEASE COHORTS
PDAC Overall (12%)
PDAC Stage I (44%)
PDAC Stage IV (3%)
T1D Managed (90%+)
Chronic Panc → T3c (70% @ 10yr)
◉ CLASS ACTION RESERVE CALCULATIONS · AMANI CLO ANALYSIS
PDAC Late Diagnosis — CA19-9 Failure Liability
RESERVE SURFACE
85% of PDAC diagnoses are late-stage (III or IV). CA19-9 has been standard of care for 30+ years with known Stage I insensitivity (<30% sensitivity). If a patient's Stage I PDAC was missed by a CA19-9 screen that the physician relied on as definitive, the standard of care question is: was there a duty to use additional screening modalities? Failure to use EUS, MRCP, or ctDNA adjuncts in high-risk patients (familial PDAC, new-onset diabetes over 50, IPMN) = potential deviation from evolving standard of care.
PRODUCT LIABILITY SURFACE · INFORMED CONSENT SURFACE
iAPS Algorithm Failure — oref0 DIY Grey Zone
BIFURCATED LIABILITY
iAPS/OpenAPS devices operated in a regulatory grey zone — community-built, not FDA-cleared as medical devices, but built on FDA-cleared CGM hardware. The oref0 algorithm is open-source and runs on thousands of real patients. If an oref0 bug (CGM dropout + insulin stack) causes hypoglycemia → seizure → harm, the liability chain is ambiguous: manufacturer (Dexcom/Omnipod)? Algorithm developers (community, no legal entity)? Hospital that advised patient on DIY loop? The 103 open issues are documented known gaps — knowledge of defects and deployment = potential negligence surface.
PRODUCT LIABILITY · ALGORITHM NEGLIGENCE
CFTR Modulator Off-Label — Trikafta for Non-CF Pancreatitis
OFF-LABEL SURFACE
Elexacaftor/tezacaftor/ivacaftor (Trikafta) is FDA-approved for CF. Emerging case series suggest benefit in CFTR-related pancreatitis (non-CF). Off-label prescribing of Trikafta at $300K/year for a pancreatitis indication without Phase 3 data creates: (1) payer liability if denied despite emerging evidence; (2) prescriber liability if harm occurs without sufficient consent disclosure; (3) manufacturer liability if off-label promotion (FDCA violation). The CFTR mutation sorry (see Adelic Pouch L2) means mechanism is partially unclear even in CF — the off-label surface is therefore broader.
OFF-LABEL RISK · PAYER LIABILITY
T1D Closed Loop Failures — CGM Dropout + Insulin Stack
SSAF-ADJACENT
The specific failure mode: CGM sensor dropout during sleep → algorithm assumes glucose is stable → basal insulin continues → glucose falling → no alert → hypoglycemia → seizure or death. This is documented in the iAPS issue tracker and the wider OpenAPS community. It is a known failure mode of the architecture (sensor data gap = algorithm blind spot). CGM dropout is not rare: sensor failures, calibration drift, compression lows. The SSAF-adjacent nature means the failure pattern resembles a smart contract exploit: a specific state (no sensor reading) triggers an unexpected algorithmic output (continued insulin delivery) with patient harm. The liability chain hits CGM manufacturer + algorithm + prescribing physician.
PATIENT HARM SURFACE · CGMD LIABILITY
◆ 4 ACTUARIAL SORRYS
SORRY-ACT-1
CA19-9 Standard of Care for 30 Years — Known Stage I Insensitivity, No Recall, No Class Action Filed
CA19-9 has been the primary pancreatic biomarker since the 1980s. Its Stage I insensitivity (<30% sensitivity) has been published in peer-reviewed literature for decades. No product liability class action has successfully challenged the standard of care around CA19-9 screening. The legal barrier: CA19-9 was never cleared for early detection (only for monitoring). If physicians used it for screening without disclosing its limitations, the informed consent surface exists — but no test case has established the duty.
VERDICT
Unlitigated known failure. Standard of care gap has been published, not actionized. First credible alternative (ctDNA panels) now in trials — when these are validated, the prior standard becomes retrospectively indefensible.
SORRY-ACT-2
iAPS/OpenAPS DIY Grey Zone — FDA Cleared Hardware, Community Algorithm, Undefined Liability
The OpenAPS/iAPS community has operated since 2015 operating under the principle that the algorithm is not a medical device (it's software on a phone), while the hardware (CGM + pump) is FDA-cleared. This grey zone has been implicitly tolerated by FDA. The 103 open issues in iAPS include documented crashes (#1882), Omnipod debug issues (#1879), and HealthKit gaps (#1884). Publication of known defects + continued use by thousands of patients + absence of formal adverse event reporting = accumulating liability surface. The grey zone closes when a patient is harmed and a lawyer reads the issue tracker.
VERDICT
Active grey zone. FDA enforcement discretion has been tacit. The first successful plaintiff argument reads the GitHub issue tracker and deposition-qualifies a developer. Bifurcated liability: CGM manufacturer (hardware) vs algorithm community (software) vs prescribing physician (endorsement).
SORRY-ACT-3
KRAS Mutation Known Since 1982 — Undruggable for 40 Years. Duty to Disclose Impossibility?
KRAS mutation was characterized in 1982. It was declared "undruggable" by the oncology community — a phrase that entered the literature and the lay press. For 40 years, PDAC patients were treated with gemcitabine + nab-paclitaxel (modest benefit, no targeted therapy) while the driver mutation was known and untargetable. Informed consent question: was there a duty to disclose that the primary driver of their cancer had no targeted therapy and was considered biochemically impossible to drug? The "undruggable" framing may have prevented patients from seeking clinical trial enrollment for KRAS-adjacent approaches. MRTX1133 Phase 1/2 (2024) changes the calculus — retroactively exposing 40 years of therapeutic nihilism.
VERDICT
Informed consent sorry. No plaintiff has argued that "undruggable" characterization = failure of informed consent. Novel theory, strong factual basis. Amani gate: requires jurisdiction analysis on whether oncology standard of care includes discussing mechanism impossibility.
SORRY-ACT-4
Teplizumab Approved 2022 — T1D Can Be Delayed. If Not Offered to At-Risk Patients, Standard of Care Sorry.
Teplizumab (Tzield, ProventionBio/Sanofi) was FDA-approved November 2022 as the first drug to delay T1D onset in at-risk individuals (stage 2 T1D, IAA/GADA/ZnT8A positive, normal glucose). It delays T1D onset by approximately 3 years in genetically at-risk patients. The standard of care question: are physicians offering T1D antibody screening (TrialNet) to at-risk first-degree relatives, and if a patient progresses to T1D without being offered teplizumab when they were at stage 2, is that a standard of care deviation? The screening infrastructure (TrialNet) exists. Teplizumab is FDA-approved. The drug costs ~$200K per course.
VERDICT
Emerging standard of care. First generation of cases where at-risk patients were not screened and not offered teplizumab will reach statute of limitations around 2025–2027. Amani gate: class action surface for pediatric T1D patients at first-degree relative risk who were not screened. Access + cost + screening gaps = multijurisdictional.
CLO GOAT GATE · AMANI JOFFE
General Counsel · EOSE Labs Inc.
GC · Scarborough Transit Connect (~$10B SSE)
GC · Scarborough Transit Connect (~$10B SSE)
CA19-9 liability surfaceMONITOR
iAPS/oref0 grey zoneHIGH RISK
Off-label CFTR surfaceMONITOR
CGM dropout harm surfaceHIGH RISK
KRAS consent sorryNOVEL THEORY
Teplizumab SOC sorryEMERGING
CLO sign-off statusPENDING
Amani note: SORRY-ACT-2 (iAPS grey zone) has the clearest near-term litigation surface. The GitHub issue tracker creates documented knowledge of defect. Recommend monitoring FDA enforcement discretion posture + any adverse event reports. iAPS community should formalize an LLC or foundation for liability shielding.
KEY DATES · ACTUARIAL TIMELINE
KRAS characterized1982
CA19-9 FDA cleared1985
OpenAPS DIY loop2015
iAPS iOS closed loop2022
Teplizumab FDA approvedNov 2022
MRTX1133 Phase 12024
PanTS NeurIPS 20252025
ctDNA panels Phase 32026–2028
CUMULATIVE EXPOSURE ESTIMATE
85% of PDAC patients diagnosed late →
~90,000 new US PDAC cases/year →
~76,500 late-stage diagnoses →
CA19-9 failure documented →
STRUCTURAL SORRY SURFACE
No single defendant. No single plaintiff class. Standard of care failure distributed across oncology practice 1985–2026. The sorry is systemic, not individual. First class action will require a jurisdiction where courts recognize systemic standard of care failure as actionable collective claim.
~90,000 new US PDAC cases/year →
~76,500 late-stage diagnoses →
CA19-9 failure documented →
STRUCTURAL SORRY SURFACE
No single defendant. No single plaintiff class. Standard of care failure distributed across oncology practice 1985–2026. The sorry is systemic, not individual. First class action will require a jurisdiction where courts recognize systemic standard of care failure as actionable collective claim.
◆ FLEET NAV