Cardiff Oncology ($CRDF) Stock Hub: Nerviano Deal Resolves Litigation, Phase 3 Funding Still Open

01 Scenario framework after the agreement

Both scenarios begin with the announced September 14 settlement. They are possibilities, not price targets or predictions.

Constructive scenario

The detailed amendment has workable economics and achievable obligations. Cardiff secures sufficient capital or a partnership, obtains any required shareholder authorisations and executes the registrational programme. Mature Phase 2 follow-up remains supportive. An increase in authorised shares would create capacity, not guarantee proceeds.

Cautious scenario

The new economics or development commitments constrain flexibility; financing is insufficient, delayed or substantially dilutive. Phase 3 timing slips or mature clinical data weaken the signal. These risks can remain after a contractual settlement and should not be replaced by the obsolete assumption that talks have yet to reach an agreement.

October 22 is the dated shareholder meeting. November 2026 is the registry’s estimated Phase 2 primary-completion month, not a promised results release. Q1 2027 remains the prior financing-dependent Phase 3 planning window; the September 14 release does not announce that enrolment has begun.

What Would Falsify This Reading

The analysis rests on Cardiff having a selected Phase 2 regimen and a settled licence dispute, while the registrational trial still depends on new financing. The following developments would materially change that reading.

• Financing that funds the registrational trial. Capital or a partnership sufficient for the roughly 640-patient Phase 3 would remove the condition the company attaches to its first-quarter 2027 start. Without it, the June 10-Q conclusion of substantial doubt about going concern remains the starting point. Source
• A different outcome at the October 22 shareholder meeting. Proposal 1 would raise authorised common stock from 150 million to 300 million shares, and the July investor warrants cannot be exercised until such an amendment is filed and accepted in Delaware. A rejection would remove the authorised-share capacity this reading assumes for future financings. Source
• Mature CRDF-004 data that change the signal. The March 18, 2026 confirmed response rate of 72.2% against 42.1% is not confirmatory evidence, and the reported PFS confidence intervals include 1.0. Mature follow-up that narrows the response difference would weaken the case; the registry lists November 2026 as the estimated primary-completion month, not a promised results release. Source Source
• Amended licence terms that constrain development. The September 14 filing announced the Nerviano settlement and amended licence, but the economics are redacted and the development obligations still matter. Disclosed royalties or Phase 3 obligations that limit flexibility would weaken the constructive reading. Source

These are observations that would weaken the interpretation, not forecasts of inevitable events.

02 NMS agreement: what changed and what is still undisclosed

The September 14 Form 8-K reports no monetary payment or admission of liability for the litigation settlement. The amended licence nevertheless includes Cardiff development/payment obligations, a patent-related fee replacing the applicable royalty, and transaction-proceeds payments to Nerviano in specified pre-Phase 3-readout assignments or changes of control. Separately, post-termination royalty tiers are payable by Nerviano to Cardiff. The amendment becomes fully operative when the litigation is dismissed with prejudice; that judicial step is not confirmed in the reviewed filing. Undisclosed percentages prevent valuation of the terms.

Cardiff and Nerviano announced a settlement and amendment to the 2017 exclusive onvansertib licence on September 14, 2026. All outstanding disputes are resolved contractually, with a full mutual release of claims. The parties plan to request dismissal with prejudice; the announcement does not establish that a dismissal order has already been entered.

The agreement announced September 14

The companies say the amendment resolves all outstanding disputes concerning worldwide onvansertib rights. It clarifies and expands the royalty structure, includes Phase 3 development objectives, gives NMS a right to appoint a board observer and to join Cardiff’s Scientific Advisory Board. A board observer is not described as an elected voting director.

No royalty percentages, payment amounts or full development-obligation schedule are disclosed in this statement. It is therefore premature to calculate the agreement’s value or describe it as economically cost-free.

Historical chronology

The original exclusive licence dates from March 2017. NMS alleged breach in February 2026, including inventorship of patents 12,144,813 and 12,263,173. Cardiff sued on May 19; NMS purported to terminate on May 27. The preliminary-injunction motion was fully briefed in July. An August 31 minute entry recorded progress at the August 28 settlement conference and scheduled a September 11 status conference. Those entries preceded the settlement announcement.

Contractual settlement and judicial procedure

The parties plan a joint request to dismiss all claims with prejudice. That is distinct from a court order already dismissing the case, and from a judgment deciding the earlier allegations on their merits. The old September 29 and October 6 briefing dates should not be presented as current binary catalysts without a later docket check.

What it means for the investment case

The unresolved licence dispute is no longer the correct current framing. The remaining questions are the amended economics, execution of the development objectives, Phase 3 funding and whether the clinical signal survives a registrational study. Settlement alone neither funds the programme nor removes clinical risk.

Historical chronology tables — superseded status, retained for context

The entries below describe the record before the September 14 agreement. References to pending motions or future briefing dates are historical and are not current catalysts.

03 What The Company Actually Is

Cardiff Oncology has one clinical asset, one lead indication, a clinical-development cost base and no product revenue. Its income statement recorded $104,000 of royalty revenue in the second quarter of 2026 and $145,000 for the half. Those royalties come from a legacy diagnostics business and are immaterial to the story in every sense except one: they are the only revenue line that exists.

The company was founded as Trovagene, a molecular diagnostics business, and repositioned into oncology therapeutics before adopting the Cardiff Oncology name. That history explains the residual royalties and it explains why the share register carries a long tail of retail holders from earlier eras.

The shape of the cost base tells a specific story about 2026. Research and development spending nearly halved between the two half-years, from $22.06 million to $12.68 million, and the company attributes that to the completion of clinical trials, to fewer patients remaining on treatment in the Phase 2 colorectal study, and to a reduction in preclinical activity as attention shifted to preparing the Phase 3. Administrative spending moved the other way, rising from $7.33 million to $9.93 million on severance agreements from the January management departures, the associated stock option modifications, and the attorney costs of the NMS dispute. A company spending less on science and more on lawyers and severance in the same period is not necessarily a company in trouble, but it is a company whose reported cost reduction is doing less work than the headline suggests.

The operating footprint

Cardiff has no commercial product or established sales infrastructure. Its principal asset is a contractual right to develop onvansertib. The September 14 amendment therefore matters substantially, but clinical execution, future royalty obligations and funding remain central to the business.

04 Onvansertib And The PLK1 Target

Polo-like kinase 1 is a serine-threonine kinase that regulates several steps of mitosis, the phase in which a cell divides. It controls entry into mitosis, the maturation of the centrosome, the assembly of the spindle that pulls chromosomes apart, and the final separation of the two daughter cells. In many tumours PLK1 is overexpressed, and cancer cells that are already under replication stress depend on it more heavily than normal cells do. That differential dependence is the therapeutic rationale: inhibit PLK1 and the tumour cell fails to complete division while the healthy cell, with more slack in the system, tolerates the same pressure better.

Onvansertib is an orally available small molecule described by the company as a highly specific PLK1 inhibitor. Specificity matters in this class because earlier attempts at mitotic inhibition produced substantial toxicity, particularly bone marrow suppression, which limited how much drug could be given alongside chemotherapy. The clinical claim Cardiff makes is not that onvansertib works alone in colorectal cancer but that it can be added to an existing chemotherapy regimen without stacking toxicity on top of it, and that the combination produces deeper and more durable tumour shrinkage than the regimen alone.

Why RAS-mutated colorectal cancer

RAS mutations, in the KRAS or NRAS genes, are present in a large share of metastatic colorectal cancers; the company describes the resulting group as a large, underserved patient population with high unmet need. Those patients are excluded from anti-EGFR antibody therapy, which does not work when the pathway is activated downstream of the receptor, so their first-line options are chemotherapy backbones combined with bevacizumab, an anti-angiogenic antibody. That is a large population with a treatment standard that has not changed fundamentally in years, which is what the company means when it describes an underserved setting with high unmet need. Preclinical work has linked KRAS-mutant cells to particular sensitivity to PLK1 inhibition, which is the biological argument for selecting this population rather than colorectal cancer generally.

The backbone question

The CRDF-004 design tested onvansertib against two different chemotherapy backbones, FOLFIRI with bevacizumab and FOLFOX with bevacizumab. The registrational regimen selected is the FOLFIRI combination, and the company has said so consistently since the End-of-Phase 2 meeting. The published headline numbers describe that arm. The selection is itself a data point: choosing one backbone out of two tested is a decision made on evidence, and the evidence behind the arm not selected has not been presented with the same prominence.

05 CRDF-004: Reading The Data Precisely

CRDF-004 is registered on ClinicalTrials.gov as NCT06106308. It is a randomised, controlled, open-label, dose-finding Phase 2 study in first-line RAS-mutated metastatic colorectal cancer, with an actual enrolment of 110 patients reported by the registry. The Form 10-Q states that the trial is being conducted in partnership with Pfizer Ignite, a fee-for-service arrangement through which a larger company supplies clinical development infrastructure; it is a services relationship, not a licensing or economic partnership on the asset. Patients received onvansertib at 20 mg or 30 mg on top of either FOLFIRI with bevacizumab or FOLFOX with bevacizumab, against control arms receiving the standard-of-care regimen alone. The primary endpoint is objective response rate; secondary endpoints include progression-free survival, duration of response, disease control rate, overall survival, adverse events and pharmacokinetics. The registry lists the study as active and no longer recruiting, with primary completion estimated for November 2026 and study completion for January 2027.

What the trial was designed to do

The company describes the primary goal as selecting the efficacious and safe dose of onvansertib plus a standard-of-care regimen for the registrational programme, and states that the trial achieved that goal. This framing is accurate and it is also a limitation that gets lost in coverage. A dose-selection study is built to choose between arms, not to prove a treatment effect against a control with statistical rigour. The arms are small, the study is open-label, and it was never powered to demonstrate superiority.

The progression-free survival numbers, and why they moved

This is the part of the record that requires care, because the reported hazard ratio has changed across data cuts and the direction of the change matters.

Two observations follow directly from the table. The January analysis used the earliest progression event from BICR or investigator assessment because events were limited. March reports BICR (0.55) and investigator assessment (0.57) separately. The numerical change from January 0.38 therefore is not a like-for-like measure of deterioration with follow-up; January 0.37 also uses pooled control arms. And the confidence intervals published for the March cut both include 1.0, the value at which there is no difference between arms. An interval running from 0.15 to 2.09 is consistent with a large benefit, with no benefit, and with harm. That is what small arms produce, and it is why the trial is a dose-selection study rather than a registrational one. The January release carried a p value below 0.05 against combined standard of care; the more recent presentation of the March cut carries confidence intervals instead, and those intervals do not exclude the null.

The response-rate separation is a useful clinical signal. It remains based on small groups and does not establish superiority in overall survival.

Durability and safety

The company reports that responses in the onvansertib arm were deeper and more durable, that four patients remained on onvansertib treatment beyond fifteen months and two beyond twenty months. At the June 23, 2026 data cut, 12 patients remained on trial: 8 in the onvansertib 20 mg or 30 mg plus FOLFIRI and bevacizumab arms and one on standard of care. Duration on treatment past twenty months in first-line metastatic colorectal cancer is a meaningful observation in itself, though a handful of long responders is not a substitute for a survival curve.

On tolerability the record is consistent across releases: the combination continued to be well tolerated with no major or unexpected toxicities and no additive adverse events observed when onvansertib was added to standard of care. Neutropenia has been described as the most common treatment-emergent event, and grade 3 or higher events as infrequent. For a mitotic inhibitor added to cytotoxic chemotherapy, the absence of additive toxicity is the technical claim the whole programme depends on.

06 The Registrational Plan And What Is Still Missing From It

Cardiff completed an End-of-Phase 2 meeting with the FDA in April 2026. That is a Type B meeting, the formal point at which a sponsor and the agency discuss whether the existing data support moving into a registrational programme and what that programme should look like. The company says it aligned with the agency on the key design elements.

What has been disclosed about the trial

The planned study is randomised and controlled, evaluating 30 mg of onvansertib in combination with FOLFIRI and bevacizumab against FOLFIRI and bevacizumab alone, as first-line therapy in patients with RAS-mutated metastatic colorectal cancer. The company has indicated an expected enrolment of approximately 640 patients across the United States, Europe and other regions, and has begun a scientific advice process with the European Medicines Agency. Initiation is planned for the first quarter of 2027.

The clause that governs everything

The August 11, 2026 company update places planned Phase 3 initiation in Q1 2027, subject to additional financing. The release does not quantify the trial budget; a numerical funding gap cannot be calculated from it.

Future funding could involve equity, partnerships or other structures. Their availability, scale, timing and dilutive effect remain uncertain until terms are announced.

What has not been published

The decisive operational questions are financing terms, a firm initiation announcement and the eventual registrational trial disclosure. Historical statements about missing registry or protocol details should not be treated as current proof that those details remain unavailable.

07 The Rest Of The Pipeline

Everything below the colorectal programme is the same molecule in other settings, and almost all of it is sponsored by academic institutions rather than by Cardiff. That structure keeps the cost low and it also means the company does not control the timelines.

The investigator-initiated programmes are exploratory and are not equivalent to the planned company-sponsored registrational programme. Their timing depends on the academic sponsors.

The antibody-drug conjugate work

At the 2026 American Association for Cancer Research annual meeting in April, the company presented preclinical data showing that onvansertib enhanced the activity of trastuzumab deruxtecan, a HER2-targeted antibody-drug conjugate, driving tumour regression and overcoming resistance in HER2-low breast cancer models. The word carrying the weight in that sentence is preclinical. These are mouse models, not patients. The scientific logic is coherent, since antibody-drug conjugates deliver a cytotoxic payload that acts on dividing cells and a mitotic inhibitor could plausibly potentiate it, and the commercial logic is obvious, since trastuzumab deruxtecan is a large and growing product. Neither makes it a clinical asset. The reviewed company update describes preclinical evidence and does not announce a human trial of that combination.

Regulatory designations

Onvansertib received FDA Fast Track designation in May 2020 for second-line KRAS-mutated mCRC. EMA orphan designation for AML dates to August 2018. These indication-specific historical designations are not marketing approval and do not automatically apply to the first-line registrational programme.

08 Second Quarter 2026 Financial Position

The numbers below come from the results release and the Form 10-Q, both filed on August 11, 2026, and they cross-check against each other and against the first quarter release.

The gap between reported expenses and cash consumption

Total operating expenses for the first half fell by $6.8 million against the prior year, from $29.4 million to $22.6 million. Net cash used in operating activities moved the other way, rising by $3.0 million from $21.1 million to $24.1 million. The two lines diverge because expense recognition and cash payment are different events: severance obligations, accrued liabilities from the wound-down trial and legal fees are settled on their own schedule. Accounts payable, which stood at $8.1 million at December 31, 2025, had fallen to $5.5 million by March 31, and paying down payables consumes cash without appearing as an expense in the period. A reader using the expense line as a proxy for burn will understate the rate at which the balance sheet is depleting.

Balance sheet

At June 30, 2026 cash and short-term investments totalled $34.520 million and liabilities totalled $10.797 million. The balance sheet shows no financial borrowings; accounts payable, accruals and lease liabilities still represent obligations ahead of common equity.

09 Going Concern: What The Filing Says And What It Means

The Q2 Form 10-Q contains a going concern disclosure in Note 1 and repeats the substance of it in management’s discussion and analysis. The company reports $34.5 million of cash, cash equivalents and short-term investments at June 30, 2026, states that this is not sufficient to meet funding requirements for at least the twelve months following the filing, and records management’s conclusion that substantial doubt exists about the ability to continue as a going concern. The financial statements are prepared on a going concern basis, which is standard practice when management intends to continue operating.

The apparent contradiction, resolved

The statements use different framing: the June 30 balance sheet and substantial-doubt assessment in the 10-Q should be read alongside the company’s forward-looking runway estimate and the subsequent July financing. A quarter-level estimate does not establish a precise exhaustion date, and the filings do not support a deterministic reconciliation based only on the calendar.

A reader who takes only the press release headline will register an extended runway. A reader who opens the 10-Q will register a going concern conclusion. Both documents were filed on the same day by the same company.

This is not new

BDO USA, the company’s auditor since 2007, included going concern language in its report on the FY2025 financial statements, dated February 24, 2026, citing recurring losses from operations and negative operating cash flows. The Q2 2026 disclosure continues a condition that was already documented at the start of the year rather than introducing one.

What it does and does not signal

A going concern conclusion is a statement about the relationship between cash on hand and a twelve-month horizon. It is not a statement that the company is failing, that the science is wrong, or that insolvency is imminent. Many clinical-stage biotechnology companies carry the language for years and resolve it repeatedly by raising capital. It establishes a material funding uncertainty within the accounting assessment horizon. It does not specify an exact cash-exhaustion date, financing price or inevitable insolvency. Future terms depend on clinical evidence, market conditions and the financing structure available at the time.

10 Capital Structure And The Cost Of Staying Alive

The share count is the number that changed most in 2026, and the way it changed illustrates what financing looks like from this position.

The July 2026 registered direct offering

On July 14, 2026 the company entered into a securities purchase agreement, closing on July 16. Institutional investors bought 8,571,429 shares with accompanying warrants at $1.05 per unit. Insiders bought a further 721,649 units at $1.455. Gross proceeds were approximately $10.05 million. The warrants carry an exercise price of $1.31 for the institutional tranche and $1.33 for the insider tranche, and they cannot be exercised before the later of six months from issuance or the date on which the authorised share count is increased. H.C. Wainwright acted as placement agent, taking a 7% cash fee and warrants over 465,157 shares at $1.3125. The offering was made off a shelf registration statement effective since May 13, 2025.

Two structural details sit inside that transaction. Insiders paid $1.455 while institutions paid $1.05 under the disclosed transaction terms. Pricing constraints and warrant terms must be considered before interpreting that difference as an investment signal. And the warrants cannot be exercised until shareholders approve an increase in authorised shares, which the company committed to seek by filing a proxy within 45 days of closing and to keep re-proposing every 60 days until it passes or the warrants expire.

The authorised-share vote: definitive proxy filed August 26, 2026, meeting on October 22

The blanks are now filled. Cardiff filed the definitive proxy statement on August 26, 2026, followed minutes later by additional soliciting material, and the special meeting is set for October 22, 2026 at 8:00 a.m. local time at the company’s offices on Flintkote Avenue in San Diego. The board fixed the close of business on August 26, 2026 as the record date, and the notice of internet availability goes out on or about August 28. The two items are the ones trailed in the preliminary version: an amendment to the certificate of incorporation raising authorised capital stock from 170,000,000 to 320,000,000 shares and authorised common stock from 150,000,000 to 300,000,000, and a routine proposal to adjourn the meeting if the votes for the first are not there.

The proxy is explicit about why the vote exists at all: the board is submitting the amendment as required under the securities purchase agreement entered into on July 14, 2026. This is a contractual obligation attached to the financing, not a discretionary governance proposal, which is consistent with the undertaking to keep re-proposing it.

Three mechanics are worth reading before the day. Quorum is a majority of the shares entitled to vote, 38,897,625 shares on the 77,795,249 outstanding at the record date. The company states that it believes both proposals are “routine” matters, so brokers could vote uninstructed shares, but it also flags that similar proposals have been classified as non-routine elsewhere; if that happened here, uninstructed broker shares would not be voted on the amendment at all, and since approval is measured against shares present and entitled to vote, the arithmetic of the room would change. Cardiff has retained Alliance Advisors as proxy solicitation agent, to assist the solicitation process; hiring an agent does not establish management’s expectation of the result. Stockholders have no appraisal rights.

The proxy also puts numbers on the overhang that the vote would unlock. Beyond the 77,795,249 shares outstanding at August 26 sit 9,757,732 shares issuable on warrants at a weighted average exercise price of $1.31, 11,504,777 on options at a weighted average of $3.29, 7,597,207 reserved under the 2021 omnibus plan and 877 issuable on the Series A convertible preferred.

One item in the definitive proxy adds a line to the capital-structure picture. Cardiff has a universal shelf registration for up to $400 million on file and an open market sale agreement with Jefferies for an at-the-market programme of up to $150 million. Against that capacity the company has sold 132,274 shares for $0.2 million net of commission to date, which is essentially nothing: the programme exists and is barely used. An ATM at a low assumed sale price can require substantial issuance to raise a given amount. Actual capacity also depends on the authorized-share count, registration terms, market demand and applicable rules; the existence of the facility does not guarantee proceeds.

The arithmetic is the point. With 77,795,249 shares outstanding on the August 6 cover, the current 150 million ceiling already carries the July offering, the warrants attached to it and the equity plans. Doubling the ceiling to 300 million creates roughly 220 million shares of unissued headroom against a company that reported substantial doubt about its ability to continue as a going concern in the same quarter. That headroom is not an issuance and does not have to be used. It is the precondition for one, and it is also the precondition for the July investor warrants becoming exercisable, also subject to their six-month waiting period, which is why the company undertook to keep bringing the resolution back every sixty days until it passes.

Dilution, measured

Shares outstanding were 68,502,000 at June 30, 2026 and 77,795,249 on the Form 10-Q cover dated August 6, 2026. That is an increase of about 13.6% inside six weeks, for roughly $10.05 million of gross proceeds. Set against a company that consumed $24.1 million of operating cash in six months, the raise buys somewhere in the region of a quarter.

The potentially dilutive securities excluded from the loss-per-share calculation at June 30, 2026 were 11,788,535 options, 432 legacy warrants and Series A convertible preferred stock equivalent to 877 shares. That table predates the July offering, so it does not include the roughly 9.3 million new warrants or the placement agent warrants. Adding them to the outstanding count gives a fully diluted picture well above 98 million shares before any further financing.

The listing threshold

Nasdaq minimum-bid compliance is based on closing bids and its applicable observation period. Historical prices around $1.00 are a risk indicator, not proof of a current deficiency notice or of current compliance.

11 Governance: A Management Team Rebuilt In One Year

The January 27, 2026 leadership changes and subsequent appointments are corporate disclosures. Their timing alone is not evidence of accounting misconduct.

The rebuild took roughly ten weeks and is now complete.

Dr. Mohindru moved from the board into the interim role and then into the permanent one, and continues as a director. The former chief executive, Dr. Mark Erlander, resigned from the board on March 27, 2026 when the separation agreements were formalised. The severance arrangements from those departures are a visible line in the 2026 cost base, contributing to the $2.6 million rise in administrative expense.

Insider participation

Insiders bought 721,649 units in the July 2026 offering at $1.455, above the $1.05 paid by institutions. The different unit and warrant terms do not by themselves reveal the insiders’ valuation assumptions. Officers and directors as a group held 5,831,354 shares, or 8.5%, at the April 20, 2026 record date, and a substantial part of the largest individual holdings consists of exercisable options rather than purchased stock. The largest holding among sitting directors is Gary W. Pace with 1,473,350 shares, or 2.2%.

Institutional register

The April 2026 proxy ownership figures are historical and should not be presented as current positions. The separately cited BlackRock July 27, 2026 amendment reports 863,259 shares, or 1.3%, at its stated event date; dated filings do not establish today’s portfolio.

The annual meeting

Shareholders met on June 11, 2026, electing six directors, ratifying BDO USA as auditor for FY2026, approving an increase in the shares available under the 2021 incentive plan to 15,150,000, and passing the advisory vote on executive compensation. A separate meeting to increase authorised share capital, required before the July warrants become exercisable, has since been called for October 22, 2026 (DEF 14A of August 26).