Formulation Design

This is a design document. Nothing here has been formulated, made, filled, or tested by Signal. Every component below is a selection between published options, with the precedent that justifies it and the confidence that precedent supports. No entry is a measured result, and none of it substitutes for bench compatibility and stability work.

Per agents/formulation.md, Signal's declared development focus was an intranasal exosome therapeutic when this document was built, and everything below designs against exosomes as the active. That was a business decision recorded as scope, not a conclusion of the four-axis comparison in active-candidates.md — which still scores naive EVs weakest of any candidate on three of four axes, and was unaffected by anything here.

Focus change flag, 2026-09-04. active-candidates.md now records Signal's declared development focus as MSC secretome (exosome-rich conditioned medium), not the purified/naive EV candidate this document was built against. This document's excipient system below has not been reworked for that active — it remains an exosome/EV-specific design (buffer, tonicity, permeation, and preservative choices reasoned from lipid-bilayer and vesicle-membrane behaviour for a purified vesicle population), which does not apply the same way to an unpurified, exosome-rich secretome that also carries soluble growth factors and cytokines outside any vesicle membrane. Recorded here as open rather than silently carried forward: the next formulation run needs to redo this document's excipient work against the new focus candidate before any of it is relied on for secretome.

Conflicts

sgl001.signal.clinic describes the product as "a shelf-stable intranasal exosome biologic" while publishing no storage condition, no shelf life, and no formulation. Fetched 2026-09-03; the phrase appears once, in the page's meta description. Every other stability-related word — "storage", "refrigerated", "frozen", "cold chain" — is absent from the page.

"Shelf-stable" is a stability claim, not a format description, and it is the one claim on the site that this document's entire subject matter exists to support or refute. Nothing published supports it, and the class evidence runs the other way:

• −80 °C is the consensus long-term storage condition for EV preparations, and one freeze-thaw cycle costs 23–36% of particles (Ahmadian et al. 2024, already recorded in sgl-001-spec.md).
• EVs held or diluted in plain PBS lose recoverable particles within minutes, not months — Görgens et al. 2022 found "using PBS as diluent was found to result in severely reduced EV recovery rates already within minutes," and drastically reduced recovery on storage "starting already within days" at every temperature tested.
• The only excipient systems shown to hold EVs stable at or above room temperature are lyophilised ones (Trenkenschuh et al. 2022; Charoenviriyakul et al. 2018), not the aqueous 10 mL metered spray the site publishes.

This is recorded here unsoftened, per the map's standing rule. It is a newer and narrower point than the storage/shelf-life gap sgl-001-spec.md already records: that artifact noted the absence of a published storage condition, and did not catch that the site simultaneously makes an affirmative stability claim in its meta description. The spec artifact owns the product-parameter record, so this is flagged for that agent rather than edited into it here. What Signal does about its own site is the operator's call.

Scope and method

The brief requires every excipient to cite an approved intranasal product or the FDA Inactive Ingredient Database (IID) as precedent, and treats an excipient with no intranasal precedent as a regulatory problem rather than a formulation choice. Three primary sources were queried directly this run:

1. The FDA Inactive Ingredient Database, July 2026 release, downloaded in full and filtered locally. Of 9,071 rows across all routes, 132 are route NASAL. That whole 132-row list is the US nasal excipient precedent universe as FDA records it, and it is reproduced in the sections below wherever it bears on a decision.
2. The openFDA drug-label index, filtered to openfda.route:"NASAL" — 694 labels — used to find which named products actually use a given excipient and at what stated level.
3. The labels themselves for the products relied on as precedent, read directly rather than through a formulary summary.

Two structural findings came out of the method itself, before any individual excipient was chosen.

There is no approved nasal biologic in the United States. Of the nasal-route labels indexed with an application number, 74 are NDAs, 238 are ANDAs, and 278 are OTC monograph entries. Zero are BLAs. The nearest things to a precedent are peptides and small proteins approved as drugs — calcitonin-salmon (NDA 020313, 32 amino acids) and glucagon (BAQSIMI, NDA 210134, 29 amino acids). SGL-001 sits on a BLA pathway (jurisdiction-map.md), and no product on that pathway has ever been approved for this route.

The IID does not contain biologic excipients at all. FDA's own guidance says so: "Excipients used in approved Biologics License Applications (BLAs) are not entered into the IID" (Using the Inactive Ingredient Database, Guidance for Industry, CDER, July 2019 draft, footnote 4). So the 132-row nasal list is drawn entirely from small-molecule and peptide NDA/ANDA products. It is the right precedent database to check — the guidance is explicit that "if an excipient is used in approved drug products for a particular route of administration, the excipient generally is not considered new and may warrant less extensive review" — but for a vesicle-based biologic it is precedent from a neighbouring field, not the same one. Both facts are recorded because they set the ceiling on how strong any precedent claim in this document can be.

The core constraint: the EV stabiliser set has no nasal precedent

The published buffer systems that keep extracellular vesicles intact were developed for frozen bulk storage in laboratory plastics, not for a nasal dosage form. Checking each of them against the IID, by route:

Precedent strength (0 = no nasal-route precedent, 3 = strong/multiple approved nasal products) and EV stability (0 = disruptive or recovery-loss for the vesicle, 3 = strongest published stabilization evidence) — scored from the precedent and tradeoff text already in this table, no new claims added.

Every single component of the best-evidenced EV stabilisation buffer has zero nasal precedent. Not one of the five appears in the nasal IID. Three of them (sucrose, poloxamer 188, albumin) have substantial precedent by other routes, which helps a safety argument but does not supply the route-specific prior use FDA's guidance describes. HEPES has no US excipient precedent by any route.

This is the single most consequential finding of this run, and it is not a gap that can be closed by choosing more carefully. It means one of three things has to happen, and the choice is a development-strategy decision, not a formulation one:

• Qualify the excipients for the nasal route. Trehalose in particular is well characterised, orally ubiquitous, and has parenteral precedent; a nasal-route safety qualification is a real but tractable package. It is also work, cost, and time that a formulation with only precedented excipients would not incur.
• Substitute within the nasal-precedented set and accept whatever stability that costs. Section Stabilisers below identifies the only two plausible substitutes.
• Move the stabiliser out of the dosage form by lyophilising, so the sugar is in a cake reconstituted at the point of use rather than in a standing aqueous nasal spray. This changes the presentation decision, which is presentation-and-device.md's territory and does not exist yet.

Buffer and target pH

The nasal mucosa's actual pH is higher than the formulation literature usually says, and that is good news here. Secondary reviews of nasal delivery routinely give the tolerated range as 5.0–6.5. The primary in-vivo measurement does not say that. Washington et al. 2000 (Int J Pharm 198:139–146, PMID 10767563) placed miniature pH electrodes 3 cm apart in the nasal cavities of twelve healthy volunteers and recorded a 30-minute stabilised baseline: anterior nasal pH 6.40 (+0.11, −0.15 SD), posterior 6.27 (+0.13, −0.18 SD), with considerable inter- and intra-subject variation. The same study sprayed 100 µL of buffered isotonic solutions and found nasal anterior pH could be pulled down only by buffers at 0.13 M and above — i.e. a low-molarity buffer does not override the nose.

The active wants neutral. Trenkenschuh et al. 2022 found "less aggregation and/or vesicle fusion occurred at neutral pH compared to slightly acidic or alkaline pH," and specifically that potassium phosphate buffer outperformed sodium phosphate buffer and PBS for EV colloidal stability on freeze-thaw.

Those two do not actually conflict, which is worth stating plainly rather than dramatising. The brief asks what to do when the mucosa's tolerated range and the active's stability optimum disagree. Here they nearly meet: measured mucosal pH is 6.3–6.4, the active prefers neutral, and the gap is under a pH unit. It is also already bridged by approved practice — SPRIX (ketorolac tromethamine nasal spray, NDA 022382) is formulated at pH 7.2, stated on its own label. An approved nasal spray sits at pH 7.2 today.

Proposed: 10 mM potassium phosphate, target pH 7.0–7.2.

Precedent strength and EV stability left blank ("—") for Molarity and Target pH — those are formulation parameters, not excipients, so the two scores don't apply.

Tonicity and osmolality

Target 270–330 mOsm/kg, taken directly from an approved label rather than from a textbook: TOSYMRA states its osmolality is "between 270 to 330 mOsmol." That is the same actuation volume (100 µL) as SGL-001's published dose figure, which makes it the closest available anchor.

Tonicity agent: sodium chloride, the most heavily precedented nasal excipient there is (IID nasal: SOLUTION 330 mg, LIQUID 7.4 mg/mL, plus SPRAY and SPRAY METERED entries; 224 of 694 nasal labels mention it; used in both TOSYMRA and calcitonin-salmon nasal spray). Confidence moderate. Precedent strength 3/3 (TOSYMRA and calcitonin-salmon both named). EV stability 1/3 — the destabiliser question raised in the second caution below is open, not resolved either way.

Two cautions that are not resolved here:

• The stabiliser is also a tonicity contributor. 25 mM trehalose adds roughly 25 mOsm; 5% sucrose adds roughly 150 mOsm. Whatever stabiliser survives the precedent problem above has to be costed into the osmolality budget before the NaCl level is fixed. Fixing an NaCl number now would be filling in a plausible figure ahead of the decision it depends on, so it is recorded open.
• Ionic strength is not neutral for vesicles. Sodium chloride is the precedented tonicity agent, but Görgens' PBS findings are about a salt-buffered system, and Trenkenschuh's preference for potassium over sodium phosphate suggests counter-ion identity matters more than a tonicity calculation captures. Whether NaCl at isotonic levels is itself a destabiliser for this preparation is an open bench question.

Viscosity and mucoadhesion

The nasal-precedented viscosity and mucoadhesive agents, from the IID:

Scores added only for Hypromellose 2910 (5 mPa·s) — the one agent from this table tracked in data/formulation.json. The other five agents were not carried into that dataset, so scoring them here would be a new judgment this run isn't making; left blank rather than guessed.

Recommendation: no viscosity or mucoadhesive agent in the initial design. Recorded as a deliberate choice, not an omission, on three grounds:

1. The trade-off the brief asks about is real and cuts against inclusion here. Increasing viscosity coarsens the plume and shifts droplet size upward. For a conventional nasal product that is acceptable or even desirable — the EMA guideline wants "the vast majority of the particles / droplets ... larger than 10 microns" to keep deposition nasal rather than pulmonary. But SGL-001's whole premise is olfactory-region deposition, and sgl-001-spec.md already records the primary finding that standard spray pumps put under 4.6% of dose in the olfactory region (Xi et al. 2016). Anything that further coarsens the plume attacks the one deposition number the product cannot afford to lose.
2. No data exists on cellulose-polymer/EV compatibility. Adsorption of vesicles onto a polymer network is exactly the loss mechanism Görgens found for plastic surfaces. Whether CMC or hypromellose does the same thing is unstudied as far as this search found.
3. It is an additive, not a subtractive, decision. A mucoadhesive can be added later against measured residence-time data. Removing one after a deposition study has been run on it is more expensive.

Confidence low — this is a reasoned deferral, not a tested conclusion. Every one of the six agents above stays available with real nasal precedent if residence time turns out to be the binding constraint.

Permeation strategy — and why the one approved option is disqualified

There is exactly one permeation enhancer with US nasal precedent, and it is a detergent that dissolves lipid bilayers.

The IID nasal list contains a single permeation-enhancing excipient: N-DODECYL .BETA.-D-MALTOSIDE (SPRAY), maximum potency listed as NA. Its precedent is TOSYMRA (NDA 210884), which contains 0.2% w/v DDM — marketed as Intravail — and reaches blood levels comparable to a 4 mg subcutaneous sumatriptan injection because of it.

DDM is also a standard laboratory reagent for solubilising membranes. Its critical micelle concentration is 0.17 mM. At MW 510.6, TOSYMRA's 0.2% w/v is 3.92 mM — about 23× its CMC. Above the CMC, DDM solubilises lipid bilayers by the classical three-stage mechanism, with the lamellar-to-micellar transition beginning at a detergent/phospholipid ratio of about R_sat = 1 mol/mol and completing at R_sol ≈ 1.6 mol/mol (dodecyl maltoside–lipid solubilisation studies, Biophys J 1998, PMID 9533703).

An order-of-magnitude check against SGL-001's own published target dose. At the site's stated 5×10⁹ particles per 100 µL — i.e. 5×10¹⁰ particles/mL — and taking a 100 nm vesicle with a lipid headgroup area of ~0.65 nm², a vesicle carries roughly 9×10⁴ lipids across both leaflets, giving on the order of 7 µM total lipid. Against 3.92 mM DDM that is roughly 500 mol detergent per mol lipid — some two and a half orders of magnitude past R_sol.

Every input to that estimate is approximate and the vesicle-lipid figure is a geometric estimate, not a measurement. It does not need to be precise. The margin is so large that no plausible correction to particle size, lipid packing, or particle count brings the ratio anywhere near the solubilisation threshold. DDM at its precedented nasal use level would be expected to dissolve the active.

Decision: no permeation enhancer. Confidence moderate on the exclusion — the calculation is an inference from established detergent physics and needs bench confirmation, but the direction is not in doubt, and the alternative is a formulation whose enhancer and whose active are mechanistically incompatible.

That leaves the permeation problem unsolved rather than solved, and this document should not pretend otherwise. The consequence is that whatever crosses the nasal epithelium has to do so unaided — which puts more weight, not less, on the deposition question sgl-001-spec.md records as open, since with no enhancer the delivered olfactory fraction is the entire lever.

Alternative enhancers were checked and none has nasal precedent: chitosan does not appear in the nasal IID and no nasal-route label mentions it; the only other surfactants in the nasal IID are polysorbate 80 (max 0.1 mg/mL in metered spray), polysorbate 20, and sorbitan monolaurate, all present as solubilisers at levels far below an enhancing dose — and all of them surfactants, carrying the same class question against a lipid vesicle at whatever level would actually enhance permeation.

Preservative, or a preservative-free presentation

The published presentation — a 10 mL multi-dose aqueous metered spray — forces this question, and every answer to it is bad for a lipid vesicle.

Under EMA/CHMP/QWP/49313/2005 (verified directly this run), a non-pressurised multiple-use metered dose nasal spray requires a preservative effectiveness/efficacy development study (Table 4.2.2, row t) and a preservative content release test (Table 4.5.2.1, row l) — both conditional on "if a preservative is present." Section 4.5.1.13 separately requires that "the number of actuations per container should be demonstrated to be no less than the labelled number of actuations."

The precedented preservatives are all membrane-active. The nasal IID lists benzalkonium chloride, benzyl alcohol (5 mg/mL, metered spray), chlorobutanol, phenylethyl alcohol, methyl- and propylparaben, potassium sorbate, thimerosal, and phenylmercuric acetate/nitrate. BAC dominates in practice: 535 of 694 nasal-route labels mention it.

BAC is a quaternary ammonium cationic surfactant. Its documented nasal toxicity is a primary finding, not an inference — Riechelmann et al. 2004 (Am J Rhinol 18:291–299, PMID 15586800) tested isolated human nasal epithelia from 15 donors and 0.05% BAC in 16 healthy volunteers over 8 days in a randomised double-blind crossover: "In vitro, BAC was ciliotoxic (p < 0.0001)", concluding that "BAC in concentrations used in nasal preparations is ciliotoxic." In vivo it did not change saccharin transport time (p > 0.8) but did cause nasal irritation (p = 0.001), burning (p = 0.0003), hypersecretion (p = 0.006), and persistent irritation (p < 0.01). The IID's own nasal levels — 40.46 mg/100 mL (0.04%) for solutions, 0.2 mg/mL (0.02%) for liquids — sit inside the range that study calls ciliotoxic.

For SGL-001 there is a second, independent problem on top of the mucosal one: a cationic surfactant at ~0.02% is a membrane-disrupting agent placed in continuous contact with a lipid-vesicle active for the whole in-use life of the bottle. No study of BAC against EV preparations was found in this search, so this is recorded as a class inference requiring bench work, not as a measured incompatibility — but it is the same physics as the DDM finding above, applied to an agent whose entire mode of action is membrane disruption.

Recommendation: preservative-free, and this is a presentation decision before it is a formulation one. Three approved products show the three ways it is actually done, and none of them is a 10 mL multi-dose bottle:

SPRIX is the closest structural analogue and the most instructive. It is a genuinely multi-dose, 100 µL metered, preservative-free nasal spray of a fragile-enough active, and it buys that by refrigerating the unopened bottle and discarding the opened one within 24 hours. Eight doses, one day. That is the shape of the trade a preservative-free multi-dose nasal spray actually makes.

Compare with what SGL-001 publishes: a 10 mL bottle, "multiple metered doses," and the word "shelf-stable." For reference on how much of a nasal bottle is even deliverable, the calcitonin-salmon label states a 2 mL fill yields "at least 14 doses" of 0.09 mL — 1.26 mL delivered, about 63% of fill — and a 3.7 mL fill yields 30 doses, about 73%. Applied to a 10 mL fill at 100 µL per actuation that implies roughly 63–73 deliverable doses, not 100 — which is consistent with, and independent primary support for, the arithmetic sgl-001-spec.md used to retire the site's withdrawn "100 doses" figure.

Confidence moderate on preservative-free as the right call; none on the in-use period, which cannot be set without stability data.

Stabilisers, cryoprotectants and bulking agents

This is where the precedent problem in the section above becomes a concrete choice. Two positions, both honest, and this run does not have the data to pick between them:

Position A — formulate on the evidence, qualify the excipient. Use trehalose. It has the deepest EV-specific evidence base of any excipient considered here: Bosch et al. 2016 (Sci Rep 6:36162) found 25 mM trehalose in the isolation and storage buffer "narrows the particle size distribution and increases the number of individual particles per microgram of protein", with significantly reduced mean particle size (p = 0.0005) and freeze-thaw protection across four cycles; Görgens et al. 2022 (J Extracell Vesicles 11:e12238) made it a component of PBS-HAT, the lead buffer of a two-year comparison, with stability "throughout several freeze-thaw cycles"; Charoenviriyakul et al. 2018 (Int J Pharm 553:1–7) found "lyophilization without cryoprotectant resulted in the aggregation" of exosomes "while the addition of trehalose ... prevented aggregation," with retained activity for about 4 weeks at 25 °C. The cost is a nasal-route safety qualification for an excipient with intradermal and subcutaneous IID precedent only.

Position B — formulate inside the precedent set. The nasal IID contains exactly two candidates with any stabiliser function:

• Mannitol — MANNITOL | SPRAY | 41.5 mg. The only sugar or polyol with nasal precedent and a stated maximum potency. A standard lyophilisation bulking agent, but a crystallising one, which makes it a weaker lyoprotectant for a membrane-bearing particle than an amorphous disaccharide like trehalose or sucrose. No EV-specific data was found. Precedent strength 2/3 (real IID nasal entry, no named marketed product). EV stability 1/3 (no EV-specific data found).
• Polysorbate 80 — POLYSORBATE 80 | SPRAY, METERED | 0.1 mg/1 mL (0.01% w/v). The nasal-precedented stand-in for the poloxamer 188 that Trenkenschuh found most effective. But polysorbate 80's CMC is in the low micromolar range, so even 0.01% is above it, which is precisely the property that made DDM disqualifying — at a hundredth the concentration and with a different head group, but not obviously safe for a vesicle. This needs measurement, not reasoning. Precedent strength 2/3 (real IID nasal entry, no named marketed product). EV stability 1/3 (unmeasured against these vesicles; shares DDM's disqualifying property but not confirmed disruptive).

Recorded as open, with a stated preference. Position A is better supported by the evidence that actually exists about this active, and excipient qualification is a known, bounded regulatory activity. Position B is cheaper and faster and rests on nothing EV-specific at all. Choosing Position B to avoid a qualification package, and then discovering mannitol does not protect the vesicles, is the more expensive mistake. But this document is not going to assert 25 mM trehalose as the answer when the decision that governs it — aqueous spray versus lyophilised cake, which determines whether the sugar is a cryoprotectant or just an osmolality contributor — has not been made.

Confidence none on a final stabiliser selection.

Excipients explicitly ruled out

Precedent strength and EV stability scored only for the five excipients tracked in data/formulation.json (DDM, BAC, Plain PBS, Citrate buffer, Chitosan). EDTA, ethanol, and phenylmercuric acetate/nitrate/thimerosal aren't in that dataset, so left blank rather than newly scored.

The design, assembled

Nothing below has been made. Confidence key follows sgl-001-spec.md: high = direct evidence for this product; moderate = evidence for the class or a hard regulatory constraint; low = derived or inferred; none = no basis found.

What needs a formulation scientist

Laboratory questions, recorded as design intent with verification marked required, per the brief:

1. DDM and BAC compatibility with the actual preparation. Both exclusions above are physics-based inferences. A simple NTA particle-count and size-distribution readout before and after exposure would confirm or refute each in a day, and would convert the two strongest recommendations in this document from reasoned to measured.
2. Polysorbate 80 at 0.01% against the vesicles. The same test. This one matters more than it looks, because polysorbate 80 is the only nasal-precedented surfactant available for surface-adsorption control.
3. Cellulose-polymer adsorption. Whether CMC-Na or hypromellose strips vesicles from solution the way plastic surfaces do (Görgens et al. 2022). If they do, the mucoadhesive option is closed permanently, not deferred.
4. Buffer screen at the product's actual concentration. Potassium phosphate vs. sodium phosphate vs. PBS, at 5×10¹⁰ particles/mL — a concentration two orders of magnitude above anything in the cited stability literature, where aggregation behaviour is least predictable.
5. pH stability window. "Neutral is better than acidic or alkaline" (Trenkenschuh) is a direction, not a specification. The width of the acceptable window around pH 7.0–7.2 has to be measured before a release specification can be written.
6. Device–formulation interaction. Whether shear through a metered pump orifice damages vesicles at all is unaddressed here and unaddressed in the literature found. It belongs to presentation-and-device.md, and it could invalidate the entire spray format independently of anything in this document.

What this run did not do

• presentation-and-device.md still does not exist as a standalone artifact, and the device specification was not attempted. A liquid/lyophilised/frozen format comparison was carried out, however — tracked in data/formulation.json's presentation key, not yet written up as prose here — and its own verdict already notes that the device, fill, and reconstitution spec remain outstanding. The findings behind that comparison are recorded so the next run, and the eventual prose write-up, start from them rather than from scratch: TOSYMRA and ENBUMYST as room-temperature unit-dose precedents; SPRIX as the multi-dose preservative-free precedent with a 24-hour in-use limit; and Trenkenschuh et al. 2022's lyophilisate holding particle size and concentration for 6 months at 40 °C. That last is the only published route to anything resembling "shelf-stable" for this active, and it is a lyophilisate, not a spray.
• No excipient supplier, grade, or compendial status was checked. USP-NF vs. multicompendial grade, and whether a supplier will provide a parenteral-grade material with a DMF, is a real gating question not addressed here.
• No extractables or leachables assessment. Required by EMA/CHMP/QWP/49313/2005 §4.5.1.11 and unaddressed.
• The IID's NA and blank maximum-potency fields were not chased. Several nasal entries relied on above (the MCC/CMC-Na co-processed entry, hypromellose 4000 mPa·s, DDM) carry no maximum potency in the database. Precedent of use exists; a precedented level does not, and for those excipients the level would have to be justified independently.

Sources

Primary sources fetched or downloaded directly this run:

• FDA Inactive Ingredient Database, July 2026 release — downloaded in full (IIR_OCOMM.csv, 9,071 rows) and filtered locally to the 132 route-NASAL entries
• Using the Inactive Ingredient Database, Guidance for Industry, CDER, July 2019 (draft) — for the route-precedent principle and the BLA-exclusion footnote
• openFDA drug label API, openfda.route:"NASAL" — 694 labels; counts by excipient and by application type
• TOSYMRA (sumatriptan) nasal spray label, NDA 210884
• Calcitonin salmon nasal spray label (Miacalcin, NDA 020313)
• SPRIX (ketorolac tromethamine) nasal spray label, NDA 022382 — via openFDA label record
• ENBUMYST (bumetanide) nasal spray label, NDA 219500 — via openFDA label record
• BAQSIMI (glucagon) nasal powder label, NDA 210134 — via openFDA label record
• EMEA/CHMP/QWP/49313/2005 Corr, Guideline on the Pharmaceutical Quality of Inhalation and Nasal Products — Tables 4.2.2 and 4.5.2.1, §4.5.1.11, §4.5.1.13, §4.2.1.20, read directly
• sgl001.signal.clinic, fetched 2026-09-03

Primary literature (abstracts and, where accessible, full text):

• Washington N, Steele RJ, Jackson SJ, et al. "Determination of baseline human nasal pH and the effect of intranasally administered buffers." Int J Pharm 2000;198:139–146. PMID 10767563
• Riechelmann H, Deutschle T, Stuhlmiller A, Gronau S, Bürner H. "Nasal toxicity of benzalkonium chloride." Am J Rhinol 2004;18:291–299. PMID 15586800
• Görgens A, Corso G, Hagey DW, et al. "Identification of storage conditions stabilizing extracellular vesicles preparations." J Extracell Vesicles 2022;11:e12238. PMC9206228
• Trenkenschuh E, Richter M, Heinrich E, Koch M, Fuhrmann G, Friess W. "Enhancing the stabilization potential of lyophilization for extracellular vesicles." Adv Healthc Mater 2022;11:e2100538. PMID 34310074
• Charoenviriyakul C, Takahashi Y, Nishikawa M, Takakura Y. "Preservation of exosomes at room temperature using lyophilization." Int J Pharm 2018;553:1–7. PMID 30316791
• Bosch S, de Beaurepaire L, Allard M, et al. "Trehalose prevents aggregation of exosomes and cryodamage." Sci Rep 2016;6:36162. PMC5099918
• Dodecyl maltoside–lipid solubilisation and the R_sat/R_sol transition: Biophys J 1998, PMID 9533703

Provenance

Input status. This document was built this run from the primary sources listed above. The IID and the EMA guideline were downloaded and parsed locally rather than read through a summary. The five product labels were read as labels. sgl001.signal.clinic was fetched directly on 2026-09-03 for the Conflicts section rather than relied on from sgl-001-spec.md's 2026-09-01 check.

Not independently re-verified this run. Two figures are carried in from sgl-001-spec.md and were not re-fetched from their own primary sources: the Ahmadian et al. 2024 freeze-thaw loss figures (23–36% per cycle) and the Xi et al. 2016 olfactory deposition fraction (<4.6%). Both are load-bearing here — the first for the Conflicts section, the second for the decision to omit a mucoadhesive — and both should be re-checked against their own publications if either becomes contested.

Read only as an abstract. Trenkenschuh et al. 2022 and Charoenviriyakul et al. 2018 were read as abstracts via Europe PMC, not full text. The Trenkenschuh abstract's statements about potassium versus sodium phosphate, neutral pH, and poloxamer 188 are quoted from it directly, but the underlying concentrations, EV sources, and effect sizes were not examined. The buffer recommendation in this document rests substantially on that abstract, so this matters.

Computed here, not cited. The ~500 mol DDM per mol lipid ratio and the ~7 µM lipid concentration are this run's arithmetic from the site's published target dose and standard bilayer geometry, not values from any source. The inputs are stated in the text so the calculation can be checked or discarded.

Not found rather than absent. No study of benzalkonium chloride, or of any nasal preservative, against extracellular vesicle preparations was found in this search. That is a negative search result, not evidence that no such study exists.