1. The modality shift: risk moves from the molecule to the system

Traditional small-molecule development is heavily molecule-specific. A team can validate the biology and still spend years optimizing potency, selectivity, pharmacokinetics, metabolism and toxicity. A new chemical series can behave very differently from the last one.

Oligonucleotide drugs begin to change that logic. Their target is largely defined by nucleotide sequence, while chemistry and delivery determine whether the drug reaches the right tissue with sufficient potency and safety. Once a chemical architecture works in a tissue, changing sequence can redirect the medicine toward another RNA target without rebuilding the entire platform. [1]

mRNA pushes the platform concept further. Cell-free IVT manufacturing can be standardized, and once a process is established, many mRNA sequences of similar size can be produced with relatively minor process adaptations. The bottleneck shifts toward delivery, expression, innate immune activation, CMC control and the biological context in which the encoded protein must work. [2]

Modality	What changes most between programs	Where major risk concentrates
Small molecule	Chemical structure	Medicinal chemistry, selectivity, ADME/toxicity
Oligonucleotide	Target sequence	Chemistry, tissue delivery, on/off-target biology, PK/PD
mRNA	Encoded sequence	Delivery, expression, innate immunity, CMC and platform execution
Personalized mRNA	Patient-specific encoded sequence	Delivery, expression, innate immunity, CMC and platform execution + real-time selection, manufacturing and treatment timing

Intismeran autogene (mRNA-4157/V940) is an extreme version of this shift. Every patient receives a different encoded sequence. Tumor and normal samples are analyzed, candidate neoantigens are selected against the patient’s tumor mutations and HLA type, and multiple neoantigens are placed into a single mRNA construct that must be manufactured and returned to that same patient. By 2018 Moderna described a construct encoding up to 34 neoantigens. [3]

The product, therefore, is not only the mRNA molecule. It is the system that selects, designs, manufactures, releases and clinically deploys it.

2. Ten years of de-risking: evidence, decision, next check

Date	Inflection point	Data / design	Decision logic	Ref
Jun 2016	Merck–Moderna collaboration	$200M upfront; Moderna is responsible for research, personalized GMP capacity and Phase I/II through human proof of concept. Merck keeps a later participation option.	Platform + execution risk still high. Merck finances the experiment without immediately taking full downstream risk.	[4]
2016–17	Preclinical	MC38 surrogate neoantigens; mutant vs wild-type T-cell readouts; separate 16-antigen concatemer experiment.	Shows mutation-specific immunogenicity and multiplexing feasibility. Does not validate human neoantigen selection or clinical efficacy.	[3]
Nov 2017	Phase I starts	KEYNOTE-603: resected tumors receive vaccine monotherapy; unresectable tumors receive vaccine + pembrolizumab; 0.04–1 mg dose escalation.	Moves risk into humans: personalized manufacturing, repeated dosing, safety and immunogenicity.	[5]
2019	Phase I interim	33 treated; no DLTs or treatment-related grade ≥3 AEs reported; 5 PR among 20 combination patients; neoantigen-specific CD8 responses detected.	Platform risk falls; comparative efficacy risk remains largely intact.	[6]
Jul 2019	Phase IIb starts	157 resected stage IIIB–IV melanoma patients; 2:1 randomization; vaccine 1 mg Q3W ×9 + pembrolizumab vs pembrolizumab alone; RFS primary.	First clean test of incremental clinical value above an active PD-1 standard.	[7]
Sep–Oct 2022	Merck opt-in	$250M option exercise; thereafter worldwide costs and profits/losses generally shared 50/50.	Strategic partner decides the program has crossed its internal threshold before public Phase II topline.	[8]
Dec 2022–2024	Phase II result	RFS HR 0.561; 18-mo RFS 79% vs 62%; 95% CI 0.309–1.017; two-sided p=0.053.	Strong signal, not definitive proof. Enough to justify registrational development.	[9]
Jul 2023	Phase III starts	INTerpath-001: randomized, double-blind, placebo/active-control; stage IIB–IV; 1,137 ultimately enrolled.	Same core biology, much higher evidentiary standard.	[10,12]
Jun 2026	5-year Phase II	RFS HR 0.51; DMFS HR 0.411; OS trend HR 0.471 with wide CI.	Durability strengthens confidence that the Phase II signal was not transient.	[11]
Aug 2026	Phase III topline	RFS and DMFS met at prespecified interim analysis; exact HRs not yet public.	First positive Phase III for an individualized neoantigen therapy; magnitude and OS remain key open questions.	[12]

3. Why melanoma, why adjuvant, and why RFS?

Melanoma was a deliberately favorable proof-of-concept indication. Cutaneous melanoma is often mutation-rich, giving a neoantigen-selection system a larger candidate pool. More importantly, checkpoint inhibitors had already shown that T-cell immunity could materially alter melanoma outcomes. By February 2019, FDA had approved pembrolizumab as adjuvant therapy for completely resected stage III melanoma based on KEYNOTE-054, where pembrolizumab reduced recurrence/death risk versus placebo with HR 0.57. [13]

That precedent reduced two uncertainties at once. Moderna did not have to prove that immune control mattered in melanoma, and it had an active standard-of-care comparator against which to test incremental value.

The adjuvant setting also matched the biology and logistics of a vaccine. After complete resection, the therapy is not being asked to shrink a bulky, heterogeneous tumor surrounded by an immunosuppressive microenvironment. It is being asked to eliminate or contain microscopic residual disease before recurrence becomes clinically visible. The weeks required for sequencing, design, manufacturing and release are also easier to accommodate after surgery than in rapidly progressing metastatic disease.

That makes RFS the natural endpoint. There is no measurable tumor left, so ORR is not meaningful. The question is whether and when cancer returns. RFS already had clear regulatory precedent in adjuvant melanoma. [13]

4. Why a confidence interval crossing 1 could still support a “positive” Phase II

The initial company result and the later peer-reviewed result can look contradictory if the statistical plan is ignored. Moderna and Merck reported a prespecified one-sided test in the Phase II signal-finding framework, while the Lancet paper also reported the conventional two-sided p-value and 95% confidence interval.

The peer-reviewed primary result was HR 0.561, 95% CI 0.309–1.017, two-sided p=0.053. Under a conventional two-sided 5% standard, the interval crossing 1 means the trial had not produced Phase III-level statistical certainty. [9]

But Phase II was a decision trial, not the final proof. The effect estimate was large, directionally consistent and clinically meaningful enough to move the probability of success. The correct investor reading in 2022 was not “proven”; it was “strong enough to justify the next, much larger experiment.”

5. Merck’s most important decision came before the public Phase II headline

In September 2022, Merck exercised its option and paid Moderna $250 million in October. From then on, worldwide development costs and profits or losses were generally shared 50/50. Moderna remained primarily responsible for process development and manufacturing, while Merck generally led clinical trials. [8,14]

The public Phase II topline was not announced until December. The companies have not disclosed the exact internal data package Merck reviewed, so it would be wrong to claim that Merck already knew the final result. But the timing still matters: after six years of direct exposure to the platform, Merck was willing to pay $250 million and accept half of future development cost before public investors saw HR 0.56.

The separate $125 million Merck equity investment in Moderna in 2018 should not be counted as an mRNA-4157 clinical milestone. It accompanied expansion of the broader oncology alliance into KRAS/shared-antigen vaccine programs. [15]

6. Capital followed the reduction in risk

The financial sequence mirrors the clinical sequence. In 2016 Merck committed $200 million while leaving Moderna responsible for the early proof-of-concept work. In 2022 it paid another $250 million to enter the global 50/50 participation structure. After that, program spending accelerated sharply. [4,8]

Moderna reported net collaboration expenses of $184 million in 2023, $390 million in 2024 and $407 million in 2025. Merck reported R&D expenses associated with the collaboration of $218 million, $358 million and $375 million over the same years. In the first half of 2026, Merck reported $184 million of collaboration R&D expense. [14,16,17]

These amounts should not be added and labeled “the cost of INTerpath-001.” The accounting bases differ, and by then the collaboration covered multiple tumor types, process development, manufacturing and shared facilities. But the direction is clear: the partners spent the least when uncertainty was highest and committed the most only after the evidence improved.

Stage	Evidence available	Risk still open	Capital decision
2016 partnership	Platform thesis + preclinical rationale	Human execution, safety, efficacy	Merck $200M; keeps option
2019 Phase I	Manufacturing feasible; tolerable; human immune activity	Incremental efficacy vs Keytruda	Randomized Phase II
2022 Phase II / opt-in	Phase II data maturing; internal data package not disclosed; years of partner diligence	Small sample; CI wide; no registrational proof	Merck $250M + 50/50 development
2023 Phase III	Phase II efficacy + durability + platform execution	Confirmation at scale; OS; commercial execution	>1,100-patient registrational program
2026 topline	Phase III RFS + DMFS success	Magnitude, OS, filings, manufacturing economics	Regulatory/commercial phase

References

1.Khvorova A, Watts JK. The chemical evolution of oligonucleotide therapies of clinical utility. Nature Biotechnology. 2017;35:238–248. Link

2.Sahin U, Karikó K, Türeci Ö. mRNA-based therapeutics — developing a new class of drugs. Nature Reviews Drug Discovery. 2014;13:759–780. Link

3.Moderna, Inc. Form 10-K for 2018. Personalized cancer vaccine product concept, preclinical MC38 studies, and platform description. Link

4.Merck and Moderna. Strategic collaboration to advance novel mRNA-based personalized cancer vaccines with KEYTRUDA. June 29, 2016. Link

5.Moderna. First-in-human dosing for Phase 1 KEYNOTE-603 of mRNA-4157. November 15, 2017. Link

6.Burris HA et al. KEYNOTE-603 Phase 1 interim results. ASCO 2019. Link

7.ClinicalTrials.gov. KEYNOTE-942 / mRNA-4157-P201. NCT03897881. Link

8.Merck and Moderna. Merck exercises option for joint development and commercialization of mRNA-4157/V940. October 12, 2022. Link

9.Weber JS et al. Individualised neoantigen therapy mRNA-4157 plus pembrolizumab versus pembrolizumab in resected melanoma (KEYNOTE-942). Lancet. 2024. Link

10.ClinicalTrials.gov. INTerpath-001. NCT05933577. Link

11.Carlino MS et al. Individualized neoantigen therapy intismeran plus pembrolizumab in resected melanoma: 5-year update of KEYNOTE-942. ASCO 2026 / J Clin Oncol 44 (suppl 16; abstr 9500). Link

12.Merck and Moderna. Phase 3 INTerpath-001 met RFS and DMFS endpoints in completely resected stage IIB–IV melanoma. August 19, 2026. Link

13.U.S. FDA. FDA approves pembrolizumab for adjuvant treatment of melanoma. February 15, 2019. Link

14.Moderna, Inc. Form 10-K for 2025. Merck individualized neoantigen therapy collaboration and 2023–2025 net collaboration expenses. Link

15.Merck and Moderna. Expanded mRNA cancer vaccines collaboration, including KRAS/shared-antigen programs and $125M Moderna equity investment. April 2018. Link

16.Merck & Co. Form 10-K for 2025. Moderna collaboration R&D expense and shared facility costs. Link

17.Merck & Co. Form 10-Q for quarter ended June 30, 2026. Moderna collaboration R&D expense and shared facility costs. Link

18.U.S. FDA. Clinical Pharmacology Considerations for the Development of Oligonucleotide Therapeutics. Final Guidance, June 2024. Link