References for the TCE-CRS-tolerance working document

What to read, in what order, and what has actually been checked

The reading queue and source list for the CRS step-up tolerance working document, with a status marker on every entry recording whether it has been verified against its source.

Nothing here has been read. Every entry below is a characterization written from a title, an abstract, or a web search summary, and every marker is ❌. That includes the numbers the specification quotes: the 17 priming/intermediate combinations in Section 5, the CAR T covariate in Section 11, and the elranatamab priming comparison in Section 9 all come from search summaries rather than from the papers. A search summary is weaker evidence than an abstract, which is itself weaker than the paper. Treat each entry as a prediction about what its source says, and correct the entry when the source is read.

The working document is the specification, and the folder’s other documents are on the project index.

Status markers

• ❌ Not checked. No claim in the working document has been verified against this source. Where the working document characterizes it, the characterization comes from an abstract or a search result.
• ⚠️ Transcribed, unverified. A number, table or model structure in the working document came from this source but has not been checked against it.
• ✅ Checked. Verified against the source.

On a reading-queue entry, ✅ means the question posed against that entry has been answered, by reading the paper or by deciding it is out of scope, and the note on the entry says which. On a source entry below the queue, ✅ keeps the stricter meaning: the claim the specification draws from it has been checked against the source. None is ✅ yet.

Where to start

Entry 1 is the project. Sections 3 and 6 of the specification are written to be replaced by what it says, so nothing else is worth reading first and Milestone 1 is finished when entry 1 is.

• Entry 1 first, twice. One pass on the abstract, the model schematic and the parameter table; a second pass on the supplement for the equations and the estimated-against-fixed table. Budget two hours, which is longer than the one-hour default on the Reading Papers page, because the deliverable is the equations rather than the conclusion.
• Entry 3 next, and it may take fifteen minutes. It is the conference version of entry 1 and it is where the count of priming and intermediate dose combinations is expected to appear.
• Entries 2 and 4 together, as the data inventory. Entry 4 is the escalation cohorts and entry 2 is the exposure-response analysis, and between them they either supply Section 5’s table or show that it cannot be filled.
• Entry 5 only after Milestone 3. It is the external test in Section 9, behaviour 3, and reading it early risks fitting the model to a result it is supposed to predict.
• Entry 6 last, for the label regimens and the interruption rules.

Journal domains are blocked from the container these notes were written in, so every link below needs a browser.

Reading queue

1. Li et al. 2026 — epcoritamab step-up dosing, RTTE model for CRS ❌

Clinical Pharmacology & Therapeutics. doi:10.1002/cpt.70362 · Wiley

Li et al. Epcoritamab Step-Up Dosing Regimen Selection and Optimization Using Repeated Time-to-Event Modeling for Cytokine Release Syndrome Risk Mitigation.

The question. All of Section 3, and the answers go into the specification rather than into this entry. The two that decide the rest of the project: what state variable drives tolerance, and which tolerance parameters were estimated rather than fixed.

What a search summary says the paper contains, all unverified: 600 patients with aggressive and indolent non-Hodgkin lymphoma given subcutaneous epcoritamab in 28-day cycles, pooled from EPCORE NHL-1 and EPCORE NHL-3; a Grade \(\ge 2\) CRS endpoint; a stimulatory effect of exposure on the hazard with a maximum, that maximum reduced by 69.7% in the 125 of 600 patients (20.8%) with prior CAR T-cell therapy; an exposure-response analysis supporting 48 mg as the full dose; and an assessment of intravenous fluids and corticosteroids on CRS risk.

Read the supplement, not only the paper. The equations, the fixed parameters, and the reason each was fixed are the deliverable, and those live in supplementary material more often than in the main text.

2. Li et al. 2025 — optimal epcoritamab dosing regimen in large B-cell lymphoma ❌

Clinical Pharmacology & Therapeutics. doi:10.1002/cpt.3588 · PMC11993285

The population PK and exposure-response analysis behind the approved regimen, by the same first author and a year earlier.

The question. Two. Does it publish a population PK model complete enough to simulate typical concentration-time profiles per regimen — this is Version B of Section 7, and this entry decides whether Version B is possible at all. And does it report CRS by dosing occasion, or only response and full-dose safety?

Open access on PMC, so this is the cheapest of the four epcoritamab entries to check.

3. AACR 2023, abstract 2798 — the conference version of entry 1 ❌

Cancer Research 83(7_Supplement):2798. AACR

Simplifying selection and optimization of step-up dosing of subcutaneous epcoritamab to mitigate CRS risk using repeated time-to-event modeling. Authors not yet recorded; the abstract predates entry 1 by three years and covers the same modelling.

The question. The dose combinations. A search summary reports priming doses of 0.004 to 0.16 mg and intermediate doses of 0.0128 to 1.6 mg across 17 combinations, with 0.16/0.8/48 mg selected on CRS event rate. If that is what the abstract says, it answers the feasibility question halfway: 17 combinations is the variation in dosing history Section 5 needs, and the approved regimen alone does not provide it.

Then ask the harder question. Whether the CRS counts per combination are published anywhere, or only the conclusion drawn from them. A model comparison of 17 regimens with no per-regimen numerator and denominator is not data this project can fit.

4. Hutchings et al. 2021 — EPCORE NHL-1 dose escalation ❌

The Lancet 398(10306):1157–1169. Lancet · PubMed 34508654

Dose escalation of subcutaneous epcoritamab in patients with relapsed or refractory B-cell non-Hodgkin lymphoma: an open-label, phase 1/2 study. Full doses from 0.0128 to 60 mg, ten sites, four countries, no dose-limiting toxicity and 48 mg taken forward.

The question. Section 5’s table, for this study. Whether CRS is reported per dosing occasion — priming, intermediate, first full dose — or pooled over cycle 1, and whether the supplementary tables break CRS down by dose cohort at each of Grade \(\ge 1\) and Grade \(\ge 2\).

This is the feasibility question in one paper. If the escalation cohorts are reported only as pooled all-grade CRS, the public inventory is thin whatever else is found, and Section 14’s first answer is already leaning negative. Check the supplementary appendix before concluding either way.

The dose-expansion companion is Thieblemont et al., JCO 2023, which reports the selected regimen rather than the alternatives, and the two-year follow-up in Leukemia 2024.

5. Elranatamab dose optimization for CRS mitigation ❌

Clinical Pharmacology & Therapeutics 2025. PMC11933221 · ASH 2022 abstract, Blood 140(Supplement 1):7174. ASH

The external test, Section 9 behaviour 3, and the first candidate for Milestone 5.

The question. Which priming regimens were compared, and what CRS incidence followed each dose of each. A search summary reports 12 mg on day 1 and 32 mg on day 4 before 76 mg on day 8 as the selected regimen, an alternative of 4 mg and 20 mg before 76 mg, exposure-response modelling of 88 patients from MagnetisMM-1 using \(C_{\max,24h}\) against any-grade and Grade \(\ge 2\) CRS, and in MagnetisMM-3 CRS in 67 of 119 patients (56.3%) on the selected regimen with 44.5% of events after dose 1 and 20.2% after dose 2.

What would confirm behaviour 3, and what would refute it. Confirmation is the smaller 4/20 priming carrying less CRS at dose 1 and more at dose 2 than 12/32 does. Refutation is the smaller priming carrying less at both. Write down which it is before fitting anything, because the model can be made to produce either.

Read this after Milestone 3, not before. A structural model tuned to a result it was shown first has not predicted it.

6. Prescribing information, Epkinly and Elrexfio ❌

The label regimens, the step-up schedules, the premedication requirements, and the restart rules after an interruption.

The question. What the labels say to do after a missed or delayed dose. That rule is the clinical expression of tolerance decay: if a label requires repeating the step-up after a gap, the gap length it names is an implicit estimate of the tolerance recovery half-life in Section 6, contributed by whoever wrote the label. It is a sanity check on \(k_{\rm out}\) and it is free.

Sources the specification’s numbers come from

Every number the specification quotes from outside itself, and where it came from. None came from the source; all came from a search summary of the source.

What the specification borrowed

Status	Number	Used in	Source
❌	17 priming/intermediate combinations, priming 0.004–0.16 mg, intermediate 0.0128–1.6 mg	Section 5	Entry 3
❌	Prior CAR T-cell therapy reduces the maximum stimulatory effect by 69.7%, in 125 of 600 patients	Section 11	Entry 1
❌	Alternative elranatamab priming regimens were studied and published	Sections 9, 10	Entry 5
❌	Full doses of 0.0128 to 60 mg in the epcoritamab escalation	Entry 4 note	Entry 4

Candidate sources for the Section 5 inventory

Searched but not yet retrieved. The inventory is built from these, and an entry leaves this list when its rows are in the table or when it is recorded as carrying none.

Where Section 5’s rows would come from

Status	Source	Expected to hold
❌	Entry 4 and its supplementary appendix	CRS by dose cohort in the escalation phase
❌	Entry 3	The count of priming/intermediate combinations, possibly per-combination CRS
❌	Entry 2	Population PK for Version B, exposure-response for the full dose
❌	EPCORE FL-1 dosing analysis, Blood 2025	Epcoritamab with lenalidomide and rituximab, a different background therapy at the same doses
❌	EPCORE NHL-6, outpatient step-up	The approved regimen with different prophylaxis, which is a prophylaxis covariate rather than a dosing history
❌	FDA and EMA review documents for Epkinly	Per-occasion CRS tables that did not reach publication

Related, not yet placed

• The other four T-cell engagers of Section 10 — teclistamab, glofitamab, mosunetuzumab, talquetamab. No entries yet, deliberately: they are Milestone 6 and each needs the same inventory question asked of it before it earns a reading entry.
• Real-world elranatamab step-up dosing, the AMbreLA interim analysis in Blood 2025. Deviations from the label schedule in practice are natural variation in dosing interval, which is the hardest quantity to identify from trial data.
• CRS grading systems. Lee 2014 against ASTCT 2019 against CTCAE. The crs_grading_system column exists because these disagree, and how much they disagree at the Grade 1/2 boundary decides whether pooling across eras is defensible.